# Changelog Source: https://help.appy.ai/changelog Product updates, new features, and improvements to Appy. ### Improvements * **Faster, higher-quality AI video** — Cole can now generate video with ByteDance's Seedance 2.5 model, a new option alongside our existing video models. ### Integrations * **Connect Day.ai** — Violet can now connect to Day.ai and use its tools directly — the first of a new class of integrations (remote MCP servers) we can turn on quickly for services like it. ### New Features * **Manage confirmation permissions from Settings** — A new Permissions tab lets you see and control which agent actions are set to "Allow always" vs. "Ask," and every organization now gets the platform's own Allow/Deny/Allow always confirmation prompt before an agent edits shared knowledge files. * **Sharper, legible images with GPT Image 2** — A new image model built for crisp text, logos, diagrams, and transparent backgrounds is now available alongside the existing models. ### Improvements * **See your paused schedules, not just the active ones** — A toggle on the Schedules page reveals recurring schedules that are currently paused, instead of having them silently disappear. ### New Features * **GPT-6 Astra and Claude Fable 5.1 are now available as model choices** — Two new frontier models — GPT-6 Astra and Claude Fable 5.1 — are now available for your agents to use, giving you more options for balancing quality, speed, and cost. ### Improvements * **Get help by email, not just chat** — You can now reach support by email in addition to chat, so you're not stuck waiting in a chat window to get help. * **Presentations now include your images** — Generated presentations can now pull in your own images directly, instead of relying only on stock or AI-generated visuals. * **Videos can now carry burned-in captions** — Generated and edited videos can now have captions burned directly into the footage, so they're viewable anywhere without relying on a separate subtitle track. ### New Features * **Ask for help without leaving the conversation** — Type --support in any conversation with an agent (web, Slack, Teams, Telegram, or email) and our support team gets a case opened immediately with everything they need to jump in — no separate email or ticket required. ### Improvements * **Safer edits to shared knowledge files** — When an agent is about to overwrite or delete a file in your organization's shared knowledge base, it now asks you to confirm first (in Slack and Teams), instead of silently proceeding or renaming around the conflict. * **Better, more flexible AI video generation** — Video generation now supports Veo 3.1 and MiniMax H3, with more resolution options (up to 4K) and duration choices. Requests outside a model's limits are automatically adjusted to the closest supported option instead of failing. ### Improvements * **Support link works again** — The "Contact Customer Support" button, shown after repeated automation failures, now correctly opens the support widget instead of landing on a broken page. * **A warmer welcome in Microsoft Teams** — When Violet joins a Teams conversation, the intro is now a rich card matching the Slack onboarding experience, instead of plain text. * **Reach Violet from anywhere in the app** — The "Message Violet" shortcut now lives at the top of the sidebar nav on every page, not just the dashboard. ### New Features * **Connect your own integrations** — Don't see the tool you need in our integration list? Now you can connect any API we don't natively support — describe what you want to do, and it gets set up as a working connection with no wait for a deploy. ### Improvements * **More QuickBooks records you can update** — Update customers, vendors, items, and accounts directly from your agent — no more workarounds or manual edits in QuickBooks itself. * **Build sharper automation triggers** — Filters on integration triggers can now check nested fields and require multiple conditions to match on the very same item in a list (for example, "fire only when this specific custom field changes to this specific value"). Rules that used to be impossible to express are now just a filter. ### Integrations * **Reflect** — Connect Reflect to create tasks, append daily notes, create notes and links, and pull your books and graphs — right from your agent. ### Improvements * **Shared PDFs, audio, and Office files now render correctly on public share links** — PDFs, audio files, and Office documents (xlsx, docx, pptx) now display properly on public share links instead of showing "Failed to load PDF file." ### Integrations * **QuickBooks, Salesforce, and Zoho Books connect in one click** — No more hunting for base URLs, auth URLs, or token endpoints — the connect flow fills in the standard values automatically. ### Improvements * **OneDrive and SharePoint files work in Microsoft Teams** — Paste a OneDrive or SharePoint link in Teams, or pick a file from the file picker, and agents can see and work with it directly. No manual re-upload needed. * **CC and BCC on agent-sent emails** — When an agent sends an email on your behalf, you can now ask it to CC or BCC additional recipients — including reply-to addressing. * **Long tasks now continue without interruption** — When an agent runs out of context mid-task, it now hands the work off to a fresh session automatically and picks up exactly where it left off. Your task keeps running; you never see an error. * **Clearer messages when integrations hit errors** — When a connected app returns an error, agents now show a plain, actionable explanation instead of raw error text — so they can self-correct or tell you exactly what went wrong. ### New Features * **Copy artifacts to your Knowledge library** — Files and documents your agents produce can now be saved directly to your shared Knowledge library in one click, so any agent can reference them in future tasks. ### Improvements * **Read Excel files natively** — Agents can now open .xlsx and .xls files directly, no conversion needed. Drop a spreadsheet into any conversation and agents can read, analyze, and work with it right away. * **Claude Sonnet 5 available** — Anthropic's latest model is now available to your agents, bringing improved reasoning and accuracy to complex tasks. ### Improvements * **Transform videos with AI** — Pass a source video to the video generator to restyle, replace subjects, or reimagine the footage entirely. Requests use Gemini Omni Flash by default; agents can switch to Kling O3 for higher-quality results. * **Connect integrations over email, no browser required** — When you send a task via email that requires connecting a new integration, the agent now returns the OAuth or API-key link directly in its reply. The flow is the same as on web or Slack. ### New Features * **Brand — set it once, every agent knows it** — Your organization's brand — colors, fonts, logo, and voice — is now captured automatically the first time you describe your company to an agent. Every agent that produces a document, image, or design reads your brand before it starts, so deliverables come out on-brand with zero configuration. Find and update it in the Knowledge page. * **Reusable document templates** — Ask any agent to save a document layout as a reusable template — a proposal format, a report structure, a Slack message template — and it gets saved to your shared Knowledge library. Pull it up by name anytime to generate a new document from the same structure. ### Improvements * **Attach videos in Slack and on the web** — Drop an .mp4, .webm, or mov file into any conversation — both on the web or in Slack — and agents can transcribe, analyze, or build on it. Video uploads already worked in Teams; now they work everywhere. * **Loops email campaigns can include images** — When building an email campaign through the Loops integration, agents can now upload images directly from your files — no manual uploading to Loops first. ### New Features * **Agent scripts can now read and write Excel files** — Scripts running in the code-run tool can import SheetJS (xlsx) to parse spreadsheets, manipulate data, and generate Excel files. Useful for agents that process data, build reports, or work with tabular content. ### Improvements * **Edit Knowledge files without leaving the page** — Markdown files in your shared Knowledge library now have an Edit button. Click it, make changes inline, and save — fo more downloading, editing locally, and re-uploading. * **Deleting a Knowledge file now asks for confirmation** — When you ask an agent to delete a shared Knowledge file, it pauses and shows a Delete/Cancel prompt before anything is removed. On Slack and Teams you get action buttons; on web you confirm by typing. Only the person who requested the deletion can approve it. * **Scheduled agent reports now arrive as content, not a file path** — Agents that write their output to a file and deliver it via \$file now send the actual report to Slack, Teams, or e￭mail as intended, rather than the raw reference. ### New Features * **Knowledge** — Add it once. Every agent can use it. Knowledge is your organization's shared reference library — pricing sheets, product docs, FAQs, and policies that every agent in your workspace can draw on whenever a task calls for them. Upload once, edit once, and every agent references the latest version automatically. [Learn more →](/your-ai-team/agent-capabilities/knowledge) ### New Features * **Artifacts** — All agent outputs — documents, images, videos, and more — are now collected in one place. Access them from any page, filter by file type or by user, and find exactly what your agents produced without hunting through past conversations. ### New Features * **Finance Team (Sage, Audrey & Sloane)** — Three specialist AI agents now handle the full financial operation of your small business. Audrey closes the books and produces month-end financials. Sage runs cash forecasts, driver-based budgets, and monthly health snapshots. Sloane manages payroll end to end — pulling time data, surfacing anomalies, running payroll after your approval, and reconciling the bank withdrawal. Connects to QuickBooks, Ramp, Brex, Stripe, Gusto, Google Sheets, and more. ### New Features * **Slack interactive messages** — Appy agents running in Slack can now send buttons, dropdowns, and other interactive controls directly in the conversation. Click a button and the agent picks up right where it left off — no need to retype your choice. * **Undo/redo in the chat input** — Cmd+Z and Cmd+Shift+Z now work in the task chat input, making it easier to fix a message before sending. * **File uploads in the task chat** — HTML, CSS, JS, and JSON files can now be attached directly in the chat input, so agents can read and work with your code files without any workarounds. * **Screenshot tool** — Agents can now capture full-page screenshots of any public URL and deliver them as files in their response. Useful for visual QA, design reviews, and web monitoring. * **Plan switching for credit-based accounts** — Credit subscribers can now upgrade or downgrade their plan directly from the billing page without going through a new checkout flow. * **Credit usage history on the billing page** — A paginated table on the billing page now shows credits consumed per execution — date, task name, who started it, and credits used. * **Integration setup during onboarding** — New users are now offered a curated integrations picker (Gmail, Slack, Notion, Linear, and more) immediately after signing up, before billing — so your first agent can hit the ground running. * **Credit breakdown on the dashboard** — The dashboard now shows your platform credits and reward credits separately, so you always know exactly what's available and where it came from. * **Agents page filters on mobile** — The activity and agents page filters no longer squish on small screens. They wrap cleanly below the title on mobile and stay inline on desktop. ### Improvements * **SEO metrics are more reliable** — The Moz SEO tool now returns partial results instead of failing entirely when one data source is unavailable. * **Connecting integrations from the app is now smoother** — The OAuth connect dialog on the integrations page no longer gets stuck on a spinner after you authenticate. Connections register immediately and the page updates without a manual refresh. * **Execution pauses gracefully on low credits** — Instead of failing mid-task when credits run out, agents now pause and notify you via your active channel (Slack, Teams, or Telegram). Resume by replying after topping up — work in progress is preserved. * **Agents see files attached to earlier Slack messages** — When you @mention Appy in an existing Slack thread, any files shared in that thread are now automatically available to the agent at the start of the task. ### New Features * **Knowledge Base File Support** - Agents can now reference uploaded PDFs, Word docs, and plain text files as persistent knowledge. Files uploaded to `/lib/knowledge/` are automatically indexed and available across executions. * **Microsoft Teams Integration** — Appy agents are now available natively in Microsoft Teams. Install from the Teams app directory and interact with Violet and your agent team in any channel or DM. ### Improvements * **Agent Trigger Reliability** — Integration triggers (GitHub, Slack, webhooks) now retry on transient failures with exponential backoff. Trigger delivery success rate improved from 94% to 99.6%. * **Slack links render correctly inside bold text** — Bold markdown around links no longer breaks the clickable link formatting in Slack. * **Margot (Graphic Designer) agent added** — Margot, a specialist graphic designer, is now part of the built-in agent roster. * **Hamilton (Web Design) agent added** — Hamilton, a specialist website designer, is now part of the built-in agent roster. ### New Features * **Multi-Agent Coordination** — Violet can now delegate to multiple specialist agents in parallel and synthesize their outputs into a single deliverable. Tasks that previously required sequential handoffs now complete significantly faster. * **InstantDB Agent Database** — Each agent now has access to a private structured database. Agents can read, write, and query data across executions — enabling stateful workflows like CRM tracking, drip sequences, and ongoing research projects. ### Improvements * **Onboarding Flow** - First-time setup now walks through connecting your first integration, hiring your first specialist agent, and running a sample task. Median time-to-first-value dropped from 18 minutes to 6 minutes. ### New Features * **Marketing Team (Marcus, Maven & Paige)** — Your Appy.AI workspace now includes a built-in marketing department: Marcus sets your brand strategy, Paige turns it into content, and Maven makes it social. Message them in Slack the same way you'd message a colleague — no new tools, no new logins. # Pricing Source: https://help.appy.ai/pricing Shared workspace credits give your team the flexibility to pay for exactly as much AI work as you need [Appy.AI](http://Appy.ai) credits power all the work for Violet and her team. There are no add-ons, no per-seat charges, no fixed overhead for agents that aren't delivering value. Learn how many credits your team requires, and scale up when demand increases, scale back when it eases. You can always monitor your credit usage within the [Billing section of the app](/your-ai-team/administration/billing). ## Get Started for Free When you establish your team's [Appy.AI](http://Appy.AI) account, you start free with 40,000 credits (a \$100 value). Put your AI team to work with full capabilities. ## Paid Plans All Appy.AI paid plans are monthly subscriptions, with a credit refresh each month on your billing date. Change your plan as you determine what your team requires. | Plan | Monthly credits | | - | - | | \$50/mo | 20,000 | | \$75/mo | 30,000 | | \$100/mo | 40,000 | | \$200/mo | 80,000 | | \$300/mo | 125,000 | | \$400/mo | 160,000 | | \$500/mo | 200,000 | | \$750/mo | 300,000 | | \$1,000/mo | 400,000 | | \$1,500/mo | 600,000 | | \$2,000/mo | 800,000 | | \$3,000/mo | 1,200,000 | | \$4,000/mo | 1,600,000 | | \$5,000/mo | 2,000,000 | Plans can be changed anytime at [builder.appy.ai/billing](https://builder.appy.ai/billing). ## Frequently asked questions Credits represent the real cost of AI work. Credits reflect actual costs to utilize LLMs through API, which Appy subsidizes so your team gets more output per dollar than you'd get going direct. It depends on what the task involves: * **Quick tasks** (summaries, lookups, short drafts, Slack responses): \~100–500 credits * **Research reports, multi-step workflows, longer deliverables**: \~1,000–3,000 credits Your credit usage broken down by task is available at [builder.appy.ai/billing](https://builder.appy.ai/billing). Yes. Credits belong to the workspace. All team members draw from the same pool. Per-member and per-department limits are on the roadmap for a future update. Monthly plan credits reset at the start of each billing cycle. Reward credits - earned through referrals and other programs - never expire and are utilized after monthly plan credits are exhausted. Any in-progress task completes. Your workspace is notified and new tasks are paused until you upgrade to a higher plan tier or you reach the renewal of your billing cycle. To get more credits before your next renewal, upgrade to the next plan tier - you'll only owe the prorated difference at that moment, and you'll pay the full fee at the next cycle. Downgrading takes effect at the start of the next cycle. Credit usage — broken down by team member and task — is available at [builder.appy.ai/billing](https://builder.appy.ai/billing). # Administration Source: https://help.appy.ai/your-ai-team/administration Manage your account: agents, activity, artifacts, websites, members, integrations, schedules, billing, and profile settings, all in one place. The [Appy.AI web platform](https://builder.appy.ai) gives you full visibility and control over your account — task history, agent roster, connected tools, schedules, usage analytics, and billing in one place. Credits, schedules, integrations at a glance — and quick links to connect Slack or Teams. Every recurring task your agents run automatically, in list or calendar view. Every file your agents have produced — searchable, shareable, and one click from your Knowledge library. Every site your agents have published, with a live preview and a link to the live page. Connect your tools so your agents can take action inside them. A live feed of every task your agents are running, have completed, or are waiting on. Your organization's shared reference library — upload once, every agent uses it. Your full org chart — every specialist, their role, and their department. Task volume, usage by team member, channel breakdown, agent workload, and assets produced. Introduce your team to Violet directly in Slack or Teams, and manage roles. Manage your subscription, payment method, and credit usage. Update your name, email, password, and Violet's email access settings. # Activity Source: https://help.appy.ai/your-ai-team/administration/activity A live feed of every task your agents are running, completed, or are waiting on — with a full-screen detail view for anything that involved delegation. Activity is your audit trail. Every task triggered by anyone on your team — through Slack, Teams, email, a schedule, or a webhook — appears here in real time. You can see what's running, what's done, and who asked for what. Screenshot 2026 05 28 At 10 13 24 AM ## Reading the activity feed Each card in the feed shows: * **Who requested it** — The team member who triggered the task, with their name and email. * **Task title** — A short description of what the agent was asked to do. * **Which agents were involved** — Violet and any specialists she delegated to. * **Status** — Current state of the task (see below). * **Progress** — A completion bar shows how far along a running task is. * **Source** — The channel the task came from (Slack, Teams, web, email, etc.), shown as an icon in the top-right corner of each card. ## Task statuses | Status | Meaning | | - | - | | **Running** | The agent is actively working on the task right now. | | **Done** | The task completed successfully. | | **Error** | The task hit an error and didn't complete. | | **Blocked** | The agent needs input from you before it can continue. | | **Idle** | No sub-tasks have started yet. | ## Filtering activity Use the filter tabs at the top to narrow by source: | Filter | Shows | | - | - | | **All** | Every task across all channels | | **Web** | Tasks started from the web platform | | **Slack** | Tasks initiated from Slack (shown when Slack is connected) | | **Teams** | Tasks initiated from Microsoft Teams (shown when Teams is connected) | | **Email** | Tasks triggered by email | | **Schedule** | Tasks run automatically on a schedule | | **Webhook** | Tasks triggered by an external webhook | **Only mine (admins only)** — Admins can toggle the feed down to just the tasks they personally started. Members already see only their own tasks, so this control isn't shown to them. ## Task detail view Screenshot 2026 05 28 At 10 12 20 AM Click any card to open the full-screen detail view. If the task involved delegation, you'll see a coverflow-style carousel — one slide per sub-agent Violet brought in, plus a slide for Violet herself. Each slide shows that agent's task list (or a summary, if there's nothing to list), the files it produced, and its status. Above the carousel, a stats bar shows the task count and file count for the whole run. The task detail view is useful for reviewing what an agent actually did — especially for scheduled tasks that run without you prompting them. # Agents Source: https://help.appy.ai/your-ai-team/administration/agents Your full org chart — every specialist, their role, and their department — with click-to-expand detail and collapsible department groups. The Agents page shows every agent on your team — who they are, what they do, and how they relate to each other. It's your org chart. ## Your AI org chart Violet sits at the top as Director of AI, coordinating the entire team. Below her, agents are grouped by department, with each department rendered as its own labeled grouping box on the canvas. Screenshot 2026 05 28 At 10 11 50 AM Screenshot 2026 05 28 At 3 15 38 PM **Click any agent card** to expand it in place and reveal its description — click elsewhere on the canvas, or expand a different card, to collapse it again. Only one card expands at a time. **Click a department's label** to collapse the whole group down to a header showing just the agent count, useful for focusing on one part of the chart at a time. Any parent node with agents reporting to it shows the same collapse toggle for its children. Use the **+** and **−** controls at the bottom-right of the org chart to zoom in and out. The fullscreen button expands the chart for easier navigation on large teams. ## Filtering your team At the top of the page, filter between: * **All** — Every agent, including custom ones you've created. * **Appy.AI** — The pre-built specialist agents that come with your account. * **Custom** — Agents you've hired yourself for roles specific to your business. ## Adding a new agent To hire a custom agent, chat with Appy.ai via Slack or Teams. Want to shape how an existing agent behaves? See [Customizing Your Team](/your-ai-team/customizing-your-team) for instructions, knowledge, and behavior settings. # Analytics Source: https://help.appy.ai/your-ai-team/administration/analytics Task volume, usage by team member, channel breakdown, agent workload, and assets produced — all in one dashboard. Analytics shows you how your team is using Appy.AI — task volume over time, which team members are most active, which channels and agents are doing the most work, and what your agents are producing. Screenshot 2026 05 28 At 10 13 30 AM ## Time range Use the dropdown in the top-right corner to change the time window: * **Last 7 days** * **Last 30 days** * **Last 90 days** All charts update together when you change the range. ## Tasks over time The **Tasks over time** chart plots daily (or weekly) task volume as an area chart. Switch between **Daily** and **Weekly** views using the toggle in the top-right of the chart. Use this chart to see whether agent usage is growing, spot spikes tied to specific projects, or identify quiet stretches where adoption may be lagging. ## Tasks by user The **Tasks by user** bar chart ranks every team member by the number of tasks they've initiated in the selected time window. Each row shows a profile photo alongside their task count. ## Tasks by medium The **Tasks by medium** radar chart shows how tasks are distributed across channels — web, Slack, email, schedule, and webhook. This tells you where your team naturally reaches for their agents most. ## Tasks by agent The **Tasks by agent** radial chart shows which agents handled the most tasks in the period. Agents are plotted by name with bar length proportional to task volume — useful for understanding which specialists are carrying the most workload and where there may be underutilized agents. ## Assets produced The **Assets produced** donut chart breaks down the files and outputs your agents have generated by type: * **Markdown** — Notes, research, reports, and summaries * **Document** — Word documents, PDFs, and formatted reports * **Image** — Generated or edited images * **Code** — Scripts, configuration files, and code snippets * **Other** — Any additional file types The total asset count is displayed in the center of the chart. Analytics data updates continuously. There's no need to refresh — the charts reflect activity as it happens. # Artifacts Source: https://help.appy.ai/your-ai-team/administration/artifacts Every file your agents have produced, in one searchable library — with sharing, editing, and a one-click path into Knowledge. Every file an agent produces during a task — a report, a spreadsheet, a generated image, a rewritten contract — lands in Artifacts automatically. You never have to ask an agent to save its output somewhere; it's already there, searchable, the moment the task finishes. ## Browsing the library Artifacts opens to a grid of tiles, newest first. Each tile shows a preview, the file name, and which task produced it. * **Search** — Find a file by name from the search box at the top of the page. * **File type filter** — Narrow the grid to one of seven categories: All file types, Images, Video, Documents, Spreadsheets, Decks, or Code. * **Creator filter (admins only)** — Admins can filter the library to files produced for a specific person. Members don't see this control — they're scoped to their own artifacts automatically, so a filter would have nothing to add. * **Infinite scroll** — Keep scrolling and more artifacts load in behind the scenes; there's no pagination to click through. Your filters live in the page URL, not just in memory. That means a filtered view — "show me every spreadsheet Maria's agents produced" — is a link you can copy and hand to someone else, and it opens to the exact same view. Click any tile to open it. ## The file detail view Opening a tile is where Artifacts earns its place as more than a file browser. The dialog previews the file inline — images, video, audio, PDFs, spreadsheets, Word docs, PowerPoint decks, CSVs, code, and markdown all render without downloading anything first. From here you can: * **Move between files** — Prev/next arrows step through the filtered set you opened from, without closing the dialog. * **Download** — Save the file locally. * **View source task** — Jump straight to the task that produced it, useful when you want the full context behind a file. * **Edit markdown files inline** — Markdown artifacts open with an editor built in. Make changes and **Save**, or **Cancel** to back out — if you've made changes, Appy confirms before discarding them. ### Sharing a file publicly Click **Share** to generate a public link to that one document — anyone with the link can view it, no Appy login required. Once shared, the button becomes **Public**; click it to copy the link again or to click **Make private** and revoke access. Making a file private and sharing it again later reuses the same link, so you don't have to redistribute a new URL each time you toggle it. An artifact share link and a [published website](/your-ai-team/administration/websites) are two different things. A share link publishes one document for anyone to view. A website is a full site an agent has built and published, with its own gallery on the [Websites](/your-ai-team/administration/websites) page. If you're looking to share a single report or file, use Artifacts; if you're looking for a live site your agents built, check Websites. ### Copying a file to Knowledge Click **Copy to Knowledge** to turn a one-off task output into a permanent reference every agent can draw on. A folder picker lets you choose where it lands in your [Knowledge](/your-ai-team/agent-capabilities/knowledge) library — or drop it at the root. This is the fastest way to build up your Knowledge library: an agent produces something useful mid-task, and instead of re-uploading it later, you file it on the spot. # Billing Source: https://help.appy.ai/your-ai-team/administration/billing Manage your subscription, payment method, and credit usage. The Billing page is where you manage your Appy.AI subscription, payment method, and credit usage. Your workspace gets a monthly credit allowance based on your plan, and credits are consumed as your agents work. See [Pricing](/pricing) for plan and credit details. Screenshot 2026 06 04 At 2 58 02 PM ## Usage and plan management The Billing page shows: * **Your plan and credit balance** — Current tier, credits remaining, and a **Reward credits** line for any bonus credits you've earned separately from your plan allowance. * **Execution credit usage table** — A per-execution breakdown of credit consumption: who initiated it, what the activity was, which channel it came through, and how many credits it used. Filter by time period (last 24 hours, 7 days, 30 days, 90 days, or all time) and search across entries. ### Switching plans Changing plans takes effect immediately: you're charged the new plan's price right away, and you receive its full credit allocation the same moment — not a prorated amount. Your monthly billing cycle restarts on the date you switch, so that date becomes your new anniversary. Because the cycle restarts immediately, you don't lose the credits you've already used this month, and you don't wait for your old cycle to end to get the new plan's allowance — it's available as soon as you confirm the change. ## Selecting a plan If you haven't subscribed yet, choose your monthly cost and credit allocation and proceed through a secure Stripe checkout to purchase. ## Managing your subscription Once your payment method is on file, the Billing page lets you upgrade, manage your subscription, and cancel. ## Canceling your subscription Click **Manage Subscription** and follow the cancellation steps. Your access continues through the end of the current billing period. There are no cancellation fees. ## Past-due payments If a payment fails, you'll see a banner here: **"Payment failed — your subscription is past due."** Update your payment method to restore access — agent interactions pause until it's resolved. Questions about your invoice or current usage? Reach out via the **Support** chat on your Dashboard or email [support@appy.ai](mailto:support@appy.ai). # Dashboard Source: https://help.appy.ai/your-ai-team/administration/dashboard Your account status at a glance and the fastest paths to getting your agents to work. The Dashboard is your home on the Appy.AI web platform. It gives you an at-a-glance read on your account status and surfaces the fastest paths to getting your agents to work. Screenshot 2026 05 28 At 10 11 36 AM ## At a glance Three summary tiles sit at the top of the Dashboard: * **Credits** — Your combined credit balance: your plan's monthly allowance plus any [reward credits](#reward-credits) you've earned, shown as a single total. Credits refresh on your billing cycle. * **Schedules** — The number of active scheduled tasks your agents are running automatically. * **Integrations** — How many third-party tools your agents are connected to. ### Reward credits Reward credits are bonus credits earned separately from your plan's monthly allowance — for example, through referrals. They're folded into the total shown on this tile, but if you want to see the reward balance on its own, check the [Billing](/your-ai-team/administration/billing) page, where it's broken out as its own line. ## Connect your communication channels If you haven't set up Slack or Microsoft Teams yet, the Dashboard walks you through it. Your agents live in chat — connecting a channel is what lets you talk to them. * **Connect Slack** — Install the Appy.AI app in your workspace to message agents directly in channels and DMs. * **Connect Teams** — Add Appy.AI to Microsoft Teams to run agents from any chat or channel. See [Installation](/your-ai-team/installation) for step-by-step setup. ## Quick links At the bottom of the Dashboard: * **Documentation** — Opens this help center in a new tab. * **Book office hours** — Schedule up to 30 minutes with the Appy.AI team for hands-on help. * **Contact Support** — Report a bug or share feedback directly with the team. # Integrations Source: https://help.appy.ai/your-ai-team/administration/integrations Connect, manage, and reconnect the tools your agents use — personal or team-wide — and see exactly what each connection can do. Integrations let your agents do real work inside the tools your business already uses. When an agent is connected to Gmail, it can send emails. Connected to GitHub, it can open pull requests. Connected to Google Sheets, it can read and write data. The Integrations page shows every tool currently connected to your account — organized by who connected them — and is where you add, manage, and remove connections. Screenshot 2026 05 28 At 10 14 03 AM ## Communication channels At the top of the page, the **Communication channels** section shows how your agents reach your team: * **Slack** — Connect your Slack workspace to message agents in channels and DMs. Shows `Connected` when your workspace is linked. * **Teams** — Connect Microsoft Teams to run agents from any chat or channel. These are your primary interfaces for working with your agents day-to-day. See [Installation](/your-ai-team/installation) for setup instructions. Disconnecting a channel is admin-only. ## Personal integrations Personal integrations (Your Integrations) are scoped to you — they're available to agents when you're the one initiating tasks, but aren't shared across the organization. Each row shows: * The tool's logo and name * **Connection status** — `Connected` (active) or `Pending` (authorization in progress) * For connected tools: who connected it and when (e.g., "Connected May 21, 2026 by [brian@appy.ai](mailto:brian@appy.ai)") ## Team integrations Team integrations list tools connected at the organization level. These are shared — every agent can use them on behalf of any team member. Org-wide integrations are typically set up by an admin and represent tools the whole team uses (e.g., a shared CRM, social media account, or a centralized Gmail account). ## Connecting a new integration Click **Add integration** in the page header to open a panel with a search field. Find the tool you want and click it to open the Connect dialog. The dialog includes: * **Access** — Choose **Personal** or **Team**. Team is only selectable by admins. * **Name** — Required only when the chosen scope already has a connection to that same integration (placeholder "e.g. Marketing account"). This is how you run two connections to one tool — for example, a personal HubSpot login alongside the team's shared HubSpot account. * Integration-specific configuration fields and/or an **API Key** field, for integrations that don't use OAuth. * **Connect** and **Cancel**. For OAuth integrations, clicking **Connect** opens an authorization popup. Once you approve access, the connection registers automatically and appears on this page. You can also ask Violet to connect something in chat — "Connect my Gmail account" or "Add our company's HubSpot CRM" both work, and she'll walk you through it. ## Managing a connection Click any integration card to open its connection detail page. What you see here depends on the type of connection: * **Communication channels** (Slack, Teams) show a simpler detail page — just the connection status and a **Disconnect** button, admin-only. * **Service integrations** (everything else — your CRM, email, spreadsheets, and so on) show a fuller detail page: **Status** (Active, or a "Needs attention" alert if the connection has expired), **Scope** (Personal or Team), **Auth method**, who connected it and when, and an **Available actions** list — every action Violet can take through that connection, in a searchable list so you can quickly check whether a specific capability is available before asking for it. ### Reconnecting an expired connection If a connection's authorization has expired, its card shows a **"Needs attention"** alert with a **Reconnect** button right on the detail page. Click it to run through the same authorization flow you used originally — once you approve access again, the connection returns to Active and any schedules or tasks that depend on it resume normally. ### Removing a connection From the "⋯" menu, choose **Delete integration**. A confirmation dialog warns that deleting permanently removes the connection and revokes access; for organization-scoped connections, it carries an additional warning that this affects the whole team. Non-admins see this option disabled with the tooltip "Only admins can delete team integrations." Personal integrations you connect are only available for your own tasks. To make an integration available team-wide, choose **Team** as the access level when connecting it (admins only), or ask Violet to connect it as an organization integration. # Members Source: https://help.appy.ai/your-ai-team/administration/members Bring your team onto Appy by having Violet introduce herself directly — then manage roles from one roster. The fastest way to get someone using Appy isn't sending them a login link — it's having Violet message them directly. Members is where you pick people from your Slack or Teams workspace and have Violet introduce herself to each one, and where admins manage everyone's role afterward. ## Introducing your team to Violet Choose **Slack** or **Teams** — if you have both connected, a toggle lets you switch between them. Search your workspace roster, multi-select the people you want to bring in (there's a select-all if you're onboarding everyone at once), then click **Introduce **. Here's the part that makes this different from a normal invite: clicking Introduce doesn't send an email with a signup link. **Violet DMs each person directly**, in Slack or Teams, and says hello. There's no account-creation step for them to stall on — they're already talking to Violet. Rows for people you've just introduced show **"Introducing…"** immediately while the invite reconciles, so you can see it's working without waiting on a page refresh. If neither Slack nor Teams is connected yet, this section prompts you to connect a channel first — Violet needs somewhere to send that first message. See [Dashboard](/your-ai-team/administration/dashboard) to connect one. ## Managing your roster Below the invite section, every member of your organization is listed with a role badge — **Admin** or **Member**. Admins can, for any member: * **Promote to Admin** or **Demote to Member** * **Remove** them from the organization — a confirmation dialog checks before removing access, since it's immediate These actions are admin-only. Members can see the roster and everyone's role, but the promote, demote, and remove controls only appear for admins. # Schedules Source: https://help.appy.ai/your-ai-team/administration/schedules See every scheduled task across your account, what state it's in, and why one might have stopped running — Violet creates, changes, and removes them. Schedules let your agents work without you prompting them. Set up a recurring task once — Violet runs it automatically on whatever cadence you define, delivering results right in Slack or Teams. The Schedules page is a read-only view of every scheduled task across your account. To create, modify, or remove schedules, ask Violet in chat. ## Status badges Every schedule shows one of five statuses: | Status | Meaning | | - | - | | **Active** | Running on its normal cadence. | | **Paused** | You (or an admin) paused it. Ask Violet to resume it whenever you're ready. | | **Needs reconnection** | An integration this schedule depends on lost its connection. | | **Auto-paused** | The schedule failed several times in a row, so Appy paused it automatically. | | **Out of credits** | Your organization is out of credits — this and other schedules are paused until you add more. | **Needs reconnection** and **Auto-paused** are the two statuses worth understanding before you see them, because they're what a stopped automation looks like from the outside. If a report you rely on suddenly stops arriving, one of these two is almost always why: * **Needs reconnection** means a tool the schedule depends on — a Slack connection, a CRM, a shared inbox — had its authorization expire. The schedule can't run until that connection is restored. Ask Violet to reconnect it, and the schedule picks back up on its next scheduled run. * **Auto-paused** means the schedule tried and failed repeatedly, so Appy stopped retrying rather than fail silently forever. Check the run history in the schedule's detail view to see what went wrong, then ask Violet to resume it once the underlying issue is fixed. Both statuses exist so a schedule that stops working tells you why instead of just going quiet. ## List view Screenshot 2026 05 28 At 10 13 45 AM The default **List** view shows each schedule as a card with: * **Task name** — What the agent is asked to do when the schedule fires. * **Recurrence** — How often it runs (e.g., "Weekly Monday at 10:00 AM Mountain"). * **Description** — A plain-language summary of the task. * **Status** — One of the five badges above. * **Last run** — Status and relative time of the most recent run. ## Calendar view Screenshot 2026 05 28 At 10 13 51 AM Switch to **Calendar** view to see the coming week of scheduled runs plotted against dates. Use the day selector at the top to navigate. Each scheduled run appears at its exact time on the correct day — click any entry to see the full task description. This is useful for spotting clustering, conflicts, or stretches where no automation is active. ## Schedule detail view Click any schedule to open its detail view: * **Next run** — When the schedule is due to fire next. * **Past 7 days** — How many times it's run in the last week, and the success rate across those runs. * **Delivers to** — Where results land (a Slack DM or channel, Teams, or "Show in app" if there's no chat delivery configured). * **Schedule** — The cadence in plain language, like "Every weekday at 9:00 AM MDT." * **Run history** — Every recent run, with its duration and, for any failed run, the error message that explains what went wrong. The run history is the first place to look when a schedule shows Auto-paused — the error messages there usually tell you exactly what to fix. ## Creating or modifying a schedule Tell Violet in chat what you want to automate and when: * "Send me a competitive intelligence update every Monday at 9 AM" * "Run the at-risk account review every week on Monday morning" * "Send a daily summary of new leads every weekday at 8 AM" * "Cancel the weekly competitive update" * "Change the Monday report to run on Fridays instead" * "Pause the daily leads summary for the next two weeks" Violet confirms the schedule and it appears on this page once active. **View more information about schedules on the [triggers page](/your-ai-team/agent-capabilities/triggers).** # Profile & Settings Source: https://help.appy.ai/your-ai-team/administration/settings Manage your personal profile, organization details, and Violet's email access settings. Your profile and account settings live in a modal accessible from your avatar at the bottom-left of the navigation. There are three tabs: **Profile**, **Organization**, and **Email**. ## Profile The Profile tab is where you manage your personal account information. **What you can update:** * **Profile photo** — Click the photo to upload a new image. * **First and last name** — Edit directly in the fields and save. * **Email addresses** — Your primary email is shown with a `Primary` badge. Add additional addresses using the input field and **Add** button. * **Password** — Click **Set password** to create or update your login password. * **Connected accounts** — Link third-party sign-in providers (e.g., Google) for one-click login. If none are connected, this section reads "No connected accounts." ## Organization The Organization tab shows your organization's name, logo, and member count, along with your public URL slug. Name and logo are admin-editable; the slug is admin-editable too. Looking to invite people or manage roles? That lives on its own page now — see [Members](/your-ai-team/administration/members). If you need to leave the organization entirely, the **Leave Organization** button is available at the bottom of this tab. ## Email The Email tab controls who can reach Violet by email, and is visible only to organization admins. **Violet's Email Address** — every organization has a dedicated inbound address for Violet, shown here with a copy-to-clipboard button. Send a message to this address and Violet processes it as a task, the same way she would a Slack DM. **Access control is an allowlist, not a blocklist.** Only mail from a sender on this list reaches Violet — everything else is rejected outright, not queued for review. If someone emails Violet and nothing happens, this is the first place to check. * **Allowed Domains** — any email from these domains is accepted (e.g., `company.com`). Add a domain and it covers every address at that domain. * **Allowed Email Addresses** — specific addresses that can email Violet, independent of domain. * **Blocked Email Addresses** — specific addresses to reject even when their domain is otherwise allowed. A block always overrides an allow, so you can permit a whole company domain and still exclude one address within it. Only organization admins can view or change the Email tab's settings. # Websites Source: https://help.appy.ai/your-ai-team/administration/websites Every site your agents have published, in one gallery — with live previews and a link to the live page. When an agent builds and publishes a website — a landing page, a microsite, a client-facing report — it shows up here. Websites is the gallery of everything your team has shipped publicly, with a live preview of each one. ## Where these sites come from An agent publishes a site to your organization's public URL when a task calls for it — ask Violet to build a page and publish it, and it appears here once it's live. Each tile is a real, currently-live page, not a draft. ## Browsing the gallery * **Live preview thumbnails** — Every tile shows a current screenshot of the actual published page. * **Search** — Find a site by name. * **Creator filter (admins only)** — Admins can filter the gallery to sites published for a specific person. Members see only their own by default, so this control isn't shown to them. * **Copy URL** — Each tile has a copy-link button for the site's public address — hand it to a client without opening the page first. * **Click through** — Click a tile's preview or title to open the live site in a new tab. A published website is not the same thing as an [artifact share link](/your-ai-team/administration/artifacts#sharing-a-file-publicly). A share link publishes a single document — a report, a spreadsheet — for anyone to view. A website is a full page or site your agents built and published in its own right. If your agent produced a document you want to share, look in [Artifacts](/your-ai-team/administration/artifacts); if it built you a page, it's here. # Agent Capabilities Source: https://help.appy.ai/your-ai-team/agent-capabilities Everything your AI team can do — from research and content to documents and connected workflows. Each specialist on your team is built on a set of underlying capabilities — the building blocks that let agents produce real work, not just responses. This page maps what's available and where to configure it. How and when your agents start working — on demand, on a schedule, by webhook, or by email. Shape any agent's behavior in plain language — instructions, tools, and skills. Share your documents, frameworks, and references so your agents speak with your perspective. Connect your agents to the 900+ tools your business already uses. Structured data your agents can read, write, and act on. Real files your agents produce — presentations, spreadsheets, Word docs, PDFs, and images you can download and share. What "Needs reconnection" or "Auto-paused" means, and how to get a schedule running again. The more context you give your agents — your preferences, your processes, your style — the better their work gets. You're not just delegating tasks, you're building a team that improves over time. # Behavior & Instructions Source: https://help.appy.ai/your-ai-team/agent-capabilities/behavior-and-instructions How each agent's behavior is shaped — instructions, tools, and skills that turn an agent into a specialist. Every agent on your team has a set of instructions that defines what it does, how it communicates, and what processes it follows. The more context you give it, the more useful it becomes. ## Instructions, tools, and skills Instructions are what make your SEO Strategist think like an SEO expert and your Sales Outreach Specialist write like a seasoned salesperson. They define each agent's approach — the questions it asks, the formats it uses, and the context it keeps in mind. Beyond instructions, each agent comes with built-in tools — capabilities like web search, file creation, and data handling that let it produce real work, not just responses. For complex, repeatable processes, agents can also have skills: specialized routines like "analyze a business proposal" or "create a content calendar." Shape any agent's behavior just by talking to it: * "Our brand voice should always be warm and approachable" * "When writing outreach emails, always mention our free trial" * "Keep blog posts under 800 words unless I say otherwise" * "Reference the customer's industry when making recommendations" Agents remember these preferences and apply them going forward. The more context you share, the better the results. ## How Appy chooses your agents' models Appy runs different agents on different models, chosen to optimize output quality while minimizing cost. A research-heavy specialist and a quick-turnaround writer don't need the same model underneath them, so we don't give them the same one — we match each agent to the model that gets the best result for what it does. We use frontier-generation models from providers like Anthropic, OpenAI, and others, and we update those choices as better options become available. Model selection isn't something you configure. That's by design: matching model to task is work the platform does so you don't have to think about it. # Automation Warnings Source: https://help.appy.ai/your-ai-team/agent-capabilities/connection-health What it means when a schedule shows Needs reconnection or Auto-paused, and how to get it running again. If a report stopped arriving, a schedule shows something other than **Active**, or a connected app suddenly seems disconnected, this page explains why — and how to fix it in under a minute. ## The short version Appy automatically pauses an automation when it can no longer run reliably, and tells you why. You never come back to a silently broken schedule. The two most common reasons: * **An integration lost its connection** — a token expired, or access was revoked upstream. The schedule shows **Needs reconnection**. * **The automation failed repeatedly** — the last two runs in a row both failed. The schedule shows **Auto-paused**. Either way, you'll hear about it — by email, and by direct message in Slack or Teams if you're connected there — and the fix is the same: **ask Violet to reconnect the integration** (or tell her what changed), and the schedule picks back up on its normal cadence. Nothing else needs to change. ## Why this happens Appy would rather stop an automation and tell you than let it run on bad data or fail silently in the background. Two conditions trigger an automatic pause: An integration the automation depends on — your CRM, your inbox, a connected spreadsheet — lost its authorization. This can happen when a token expires, a password changes, or access is revoked from the other side. Appy won't run an automation against a broken connection, so it pauses the schedule rather than let it fail (or worse, run with stale access) on the next trigger. The automation's last two runs in a row both failed. Rather than keep retrying — and keep failing — Appy stops it and flags it for you to look at. This is usually a sign something upstream changed: a renamed column, a deleted record the automation expected, a permissions change. Check the recent runs on the [Schedules page](/your-ai-team/administration/schedules) to see what went wrong before asking Violet to resume it. Both states are visible right on the [Schedules page](/your-ai-team/administration/schedules), next to the affected automation. ## How you're notified The moment an automation is paused, its owner is notified — by email, and by direct message in Slack or Teams if you have one connected. The message says plainly what happened and what to do next. You won't discover a paused automation only by noticing a report never arrived. ## How to fix it 1. **Read the notification or the schedule's status** — it tells you which of the two states applies. 2. **For "Needs reconnection":** ask Violet to reconnect the integration, or reconnect it yourself from the [Integrations page](/your-ai-team/administration/integrations). Once the connection is live again, ask Violet to resume the schedule. 3. **For "Auto-paused":** check the recent runs on the Schedules page for the error, fix whatever caused it (often on the connected app's side), then ask Violet to resume the schedule. Once resumed, the schedule picks back up on its original cadence — nothing else needs to be reconfigured. # Databases Source: https://help.appy.ai/your-ai-team/agent-capabilities/databases Structured data your agents can read, write, and act on — customer lists, trackers, pipelines, and more. When your agents have access to structured data, they stop being just smart assistants and become real operational tools. Databases let you give agents customer lists, project trackers, sales pipelines, or any organized information your business runs on. With database access, an agent can look up specific records, add new entries, update existing ones, and use that structured data to inform its work. Here's how different businesses use databases: * **Consultants** track client projects with fields for deliverables, deadlines, and status — their agent becomes a lightweight project management system * **Coaches** monitor client goals with progress percentages and milestone dates — sessions become data-driven and outcomes trackable * **Agencies** manage content calendars with post titles, publish dates, and performance metrics — their agent orchestrates the entire content workflow You can control how much visibility you have into the data. Databases can work entirely behind the scenes, or you can view and edit records directly. Most people start without databases — they're most useful once you have structured business data you want agents to actively work with, not just reference. # Document Creation Source: https://help.appy.ai/your-ai-team/agent-capabilities/document-creation Real files your agents produce — presentations, spreadsheets, Word docs, PDFs, and images you can download and share. Your agents don't just write text in a chat — they create actual files you can download, share, and use. Ask for a deliverable in plain language and your agent produces a polished, ready-to-use file. ## Documents and files * **Presentations** — Pitch decks, meeting slides, and proposals delivered as PowerPoint files * **Spreadsheets** — Financial models, budgets, trackers, and reports as Excel files * **Word documents** — Your agents can create polished Word documents — complete with formatted headings, tables, images, and even a table of contents. Need a proposal, a report, or a client deliverable? Just describe what you need and your agent will put it together. They can even pull in charts or images from earlier in the conversation. * **PDFs** — Polished reports, one-pagers, and printable documents * **Images** — Graphics for social media, blog headers, and marketing materials Ask for "a pitch deck for our Q2 investor meeting" or "a spreadsheet tracking monthly marketing spend" and your agent delivers a ready-to-use file. # Integrations Source: https://help.appy.ai/your-ai-team/agent-capabilities/integrations Connect your agents to the 2,000+ tools your business already uses — securely, with scoped permissions. Integrations let your agents work inside your existing stack — reading data, taking actions, and interacting with other platforms on your behalf. Connect once and your agents handle the rest. Connect with over 2,000 applications — including HubSpot, Slack, Salesforce, Google Docs, Google Calendar, Gmail, Google Drive, Google Sheets, Jira, GitHub, Linear, Shopify, Stripe, Intercom, and more. For each integration, your agents get access to specific actions: reading emails, creating calendar events, updating records, sending messages. When a task requires an integration, your agent walks you through connecting your account. Authorize the service once, and your agents can work with it from then on. Disconnect any integration at any time. Your agents never see or store your passwords. Each operates within defined, scoped permissions — no blanket access. ## Connecting a service that isn't in the catalog If a tool you use isn't in the 2,000+ app catalog, just tell Violet what you want to connect and what you need it to do. Violet researches the service's API and builds the connection herself — you never touch a specification, a schema, or any technical setup. Once she's built it, she sends you one link to authorize the service. From there it works like any other integration: your agent reads data, takes actions, and asks you to confirm the first time it does anything that changes data. Two ways this can go, depending on how the service exposes itself: * **API key or token.** Violet defines the connection from the service's API docs and sends you a secure page to enter your credentials. She never asks for a key, password, or account ID in chat. * **Sign in with your account (OAuth).** Some services — Day.ai is one — only expose their capabilities through a remote MCP server behind your own sign-in page. For these, you authorize by signing in on the service's own page, the same way you would connect any app; Violet never sees or holds a credential for it. Tell Violet whether the connection should only read data or also make changes — she asks up front, and it shapes what the connection is allowed to do. Adding capability later is just asking again; the connection doesn't need to be rebuilt from scratch. ### Who can see and use it Once a custom integration is connected, it's visible to everyone in your organization — not just the person who built it. Any teammate can ask their agent to use it, the same as a catalog integration. The connection you authorize is still yours: your credentials stay tied to you, and a teammate using the integration doesn't see or borrow them. If the team should share one connection instead of everyone connecting individually, that's a deliberate step — ask Violet to make it a shared, organization-wide connection. Only the person who originally built the integration can change what it does. Everyone else can use it as-is. # Knowledge Source: https://help.appy.ai/your-ai-team/agent-capabilities/knowledge Your organization's shared reference library — upload once, every agent uses it. **Add it once. Every agent can use it.** Knowledge is your organization's shared reference library — the pricing sheets, product docs, FAQs, and policies your agents need to do work that's actually about your business. Upload your documents once and every agent in your workspace can draw on them whenever a task calls for them. *** ## What is Knowledge? Knowledge is a shared file library built into your Appy.AI workspace. It works like the folders you already use: you upload documents, organize them into folders, and your agents reference them automatically when they're relevant to a task. Without Knowledge, every request starts from scratch. You attach the pricing sheet again, paste the product details again, explain the policy again. With Knowledge, your agents know what you've filed — and they use it. The Knowledge library showing organized folders and files *** ## Who can do what | Role | Can access Knowledge library | Can upload and edit files | | - | - | - | | Admin | ✅ Yes | ✅ Yes | | Team member | ✅ Yes | ✅ Yes | | Guest | ✅ Yes (read-only) | ❌ No | | Agents | Read automatically | ❌ No (see below) | **Admins and team members have identical access.** Both can view, upload, create folders, rename, move, copy, delete, and edit and save markdown files. Knowledge is a fully shared workspace resource — there's no admin-only gate on managing it. **Guests** have read-only access: they can browse, preview, and download files, but all write operations are rejected. **Agents** read the library automatically but can't write to it on their own initiative — what's in there reflects what your team has filed. They can still get files into Knowledge indirectly: see [Asking Violet to save a file](#asking-violet-to-save-a-file) below. `/brand` and `/templates` (and the `brand.md` and `templates/README.md` files inside them) can't be renamed, moved, or deleted by anyone, including admins. Their contents, however, are fully editable — update your brand guidelines or template structure anytime. The Knowledge page with the brand folder and file listing *** ### Uploading documents Open your workspace and click **Knowledge** in the left sidebar. Click **+ Add** in the top right, then choose **Upload file** and select your document. The file is immediately available to every agent in your workspace. You can upload any document type — PDFs, Word files, spreadsheets, plain text. Common examples: * Pricing sheets * Product one-pagers * Customer FAQs * Policy documents * Onboarding materials * Brand guidelines ### Organizing with folders Keep your library tidy with folders: Click **+ Add** and choose **New folder**, then name it and press Enter. Drag files into the folder, or upload directly into it. You can rename, move, and delete folders at any time. ### Editing a file 1. Click any file in the Knowledge library to open it. 2. Make your changes and click **Save**. Every agent in your workspace references the updated version immediately. There's no publish step — save and it's live. ### Asking Violet to save a file If a useful document comes up mid-conversation, you can ask Violet to save it to the Knowledge library directly: > "Violet, save this to Knowledge." She'll file it for you without interrupting the task. You can also copy an artifact an agent produced straight into Knowledge from the artifact view — useful for turning agent output into a reusable reference without leaving the conversation. The Knowledge page in edit mode showing file editing ### Deleting a file Click the file to open it, then click the **delete** icon. The file is removed from the library immediately. Agents will no longer reference it. *** ## What agents use Knowledge for When an agent works on a task, it searches your Knowledge library for documents that are relevant. It pulls the reference it needs — a price list for a proposal, a policy doc for a support reply, a product sheet for a research brief — and uses it automatically. You don't need to tell the agent which document to use. It finds the right one from what you've filed. *** ## Frequently asked questions Not at a per-agent level. The entire Knowledge library is available to all agents in your workspace. Organize your library with folders to keep it clear, and only file documents you're comfortable sharing across the whole team. The moment you save an edit, every agent in your workspace references the new version. There's no delay and no re-upload needed. Most common document types are supported: PDF, DOCX, XLSX, TXT, CSV, and more. If a file type isn't supported, Appy will let you know when you try to upload it. Not on their own initiative — agents don't edit the library unless a person acts on it. But agents don't leave you stuck with manual re-uploads either: you can copy an artifact an agent produces straight into Knowledge, or simply ask Violet to save something there mid-conversation, and it's filed without you leaving the task. Deleted files cannot be recovered from Knowledge. Keep a backup copy in your own storage if the document is critical. *** ## Tips **File the documents your agents use most.** If you find yourself attaching the same file to requests frequently, it belongs in Knowledge. * **Use clear file names.** Agents search by content and file name. A file called `pricing-q3-2026.pdf` is easier to find than `final-v2-LAST.pdf`. * **Keep it current.** The power of Knowledge is that agents always reference the latest version. Review your library when your pricing or products change. * **Organize from the start.** It's easier to maintain a tidy library than to reorganize a cluttered one. # Triggers Source: https://help.appy.ai/your-ai-team/agent-capabilities/triggers Control when your agents start working — on demand, on a schedule, by webhook, or by email. Triggers control when your agents start working. By default, agents respond when you message them — but set up a trigger and they'll run on their own, 24/7, with no maintenance required. ## Trigger types * **On demand** — @mention or DM Appy.AI in Slack or Teams and your agent gets to work immediately. This is how most interactions happen. * **Scheduled** — Agents run on a recurring schedule: daily, weekly, monthly, or at a specific time. Great for morning briefings, weekly competitive reports, or monthly performance summaries. * **Integration triggers** — External tools can kick off your agents automatically. When something happens in another system — like a new lead added to your CRM — your agent handles it. * **Webhook** — External systems trigger your agent by sending an HTTP POST request. This is how you connect your agent to tools like Zapier, Make, your CRM, or a form on your own website. * **Email** — Every agent has its own email address. Send it a message and it starts working on what you asked, with the email content as its task. You can stack multiple triggers on the same agent. A reporting agent might run on a weekly schedule and also be available to message anytime for ad-hoc requests. The easiest way to set up triggers is just to ask: "Send me a competitive update every Monday morning" or "Change my daily report to 6am instead of 8am." ## Scheduled triggers Scheduled triggers turn a one-time request into a standing instruction. Instead of asking your agent every Monday, it already knows it's Monday — and shows up with the answer. Once a schedule is set, the agent will: * Wake up at the time you specified (in your timezone) * Carry out the task exactly as it would if you'd asked in chat * Deliver the result to your chosen destination (a DM, a channel, or a thread) * Repeat on the cadence you set, until you pause or remove it The schedule runs whether your computer is on or off — Appy handles it in the cloud. ### Examples * *"Every weekday at 8 AM, have Sarah send me a briefing of my calendar and any unread emails marked important."* * *"Every Monday at 9, have Scout check what our top three competitors shipped last week and post a summary to #competitive-intel."* * *"On the first of every month, have Lincoln pull an SEO audit on appy.ai and DM it to me."* * *"Every Friday at 4 PM, ask Piper to surface 15 new leads matching our ICP and drop the list in #pipeline."* You can include the destination ("DM me," "post in #channel-name"), your timezone if it differs from the default, and any context the agent needs for that specific task. To change a schedule — time, destination, or cadence — just tell Violet: "stop my Monday competitive check," "move my morning briefing to 7 AM," or "have it go to #leadership instead." ### Managing schedules Violet is your control panel for scheduled work. To pause, change, or remove a schedule, tell her in chat. To see what's currently scheduled, visit [https://builder.appy.ai/schedules](https://builder.appy.ai/schedules). > **Why route everything through Violet?** Violet has full context on your team — which agents you have, what they're good at, and how they connect to your tools. Owning the schedule flow means nothing drifts out of sync with the agent running it. ## Webhook triggers Webhook triggers let external systems kick off an agent by sending it an HTTP request — the connection point for Zapier, Make, your CRM, a form on your website, or any tool that can make a POST call. To set one up, tell the agent what should happen and what the trigger should be called — for example, "Set up a webhook trigger called `new-lead` that drafts an outreach email whenever it fires." The agent creates the trigger and gives you back a unique URL: ``` POST https://builder.appy.ai/api/trigger/webhook/{agent}/{shortId} ``` Send a POST request to that URL from your external tool. The agent runs the task you defined when you set up the trigger, and if the request includes a body, that payload is attached to the run so the agent can reference "the incoming data" without you having to template it in. Webhook trigger endpoints accept JSON, form-encoded, and multipart bodies — so a plain HTML form on your own website can post to a webhook trigger directly from the browser, no backend glue required. Endpoints also allow cross-origin requests, since the URL itself (not a cookie or API key) is the credential. ### Examples * **Marketing agency** — A new lead from a website form triggers an outreach sequence. * **Business coach** — A client completing an intake form triggers a welcome sequence. * **SaaS company** — A new trial signup triggers an onboarding email series. ## Email triggers Every agent has its own email address. Send it a message — or forward one to it — and the agent starts working, with the email's subject, body, and attachments passed along as its task. This is useful for forwarding an inbound request straight to the agent responsible for it, or for systems that can send email but can't make an HTTP call. To find Violet's address and manage who's allowed to email her, open your avatar (bottom-left of the navigation) → **Settings** → **Email**. See [Profile & Settings](/your-ai-team/administration/settings#email) for the full allowlist details — admin-only, and worth checking first if an email trigger seems to be silently ignored. ## Getting started with triggers Here's how to set up your first triggers: 1. **Enable on demand.** This should almost always be on. It lets you message your agent anytime. 2. **Identify routine tasks.** What work happens on a schedule? Set up scheduled triggers for those. 3. **Connect external tools.** If you use Zapier, Make, or other automation tools, set up a webhook trigger and connect them. 4. **Test each trigger.** Make sure each one works the way you expect before relying on it. # Customizing Your Team Source: https://help.appy.ai/your-ai-team/customizing-your-team Your team, your rules. Hire, fire, and mentor agents as your business evolves. Your AI team starts ready to work — and gets sharper the more it knows about your business. Hire new specialists, retire the ones you don't need, and coach your agents in plain language. ## Hire When you need a specialist that doesn't exist yet, just describe the role and Violet builds it. Try something like: * "I need an agent for customer onboarding emails" * "Create an agent that monitors social media mentions weekly" * "I want a specialist who can write product descriptions for our e-commerce store" No code required. Your new agent inherits all the business context you've already shared, so it starts informed. ## Fire When a role is no longer relevant, retire it. Tell Appy.AI: * "I don't want that Spring event planning agent anymore" * "Fire Paige. I want to add my own content strategist agent." Your team stays lean and focused. You can always bring new agents online when priorities shift. ## Mentor The fastest way to improve an agent is to tell it what you want — no code, no configuration. Give agents **skills** (step-by-step workflows) or **knowledge** (reference documents they draw on automatically): * "Save this brand voice document into org knowledge so all agents can reference it." * "Create a new skill for my executive assistant that helps her organize my inbox the way I like it." * "Make sure my sales agent has this annual report in their knowledge base." * "Create a new client onboarding skill — I'll walk you through the steps." Appy.AI applies these preferences going forward. The more context you share, the better the results. ## Growing your team Your AI team starts ready to work with a core roster of specialists, and grows with your business. Add specialists whenever you hit a gap in what your current team can handle. Some businesses run lean with just the core agents. Others build out dozens of specialists across functions. Your AI team scales with you. Power users can fine-tune agent behaviors through the web platform — but for most people, conversation is all you need. # Installation Source: https://help.appy.ai/your-ai-team/installation Get Appy.AI running where your team already works. Appy.AI lives inside Slack and Microsoft Teams — so your team gets a full AI workforce without switching to a new tool. When you sign up, create your account using your Microsoft or Slack login. You'll have access immediately. ## Slack Appy.AI installs in your Slack workspace as an app. Once it's there, message it like any other teammate. To get started, sign up via Slack. Authorize the app to access your workspace, and you're ready to go. Screenshot 2026 05 21 At 2 27 46 PM Open a direct message with Appy.AI and start telling Violet what you need. No commands to memorize, no special syntax — just describe your goal like you would to a colleague. Use `--` or the word `aside` in a thread to have Appy.AI ignore a message. Useful when you want to talk with teammates in a shared channel without Violet jumping in. *Example: `aside @steve are you following this conversation with Appy?`* ## Microsoft Teams Appy.AI works inside Microsoft Teams as a custom app. Once installed, your team can message AI agents without leaving Teams. Setup requires a Microsoft global administrator to upload the app and grant permissions — about 10 minutes total. Step-by-step guide to installing and connecting the Appy.AI app in your Teams workspace. ## Email Every agent also has its own email address, so email works as a third channel alongside Slack and Teams — send it a message and the agent gets to work. Find the address and manage who's allowed to send to it under your avatar → **Settings** → **Email**. See [Profile & Settings](/your-ai-team/administration/settings#email) for the full setup, including the sender allowlist. ## Web platform Log in to [builder.appy.ai](http://builder.appy.ai) to manage your account: 1. Monitor agent activity 2. Billing and account upgrades 3. Integrations 4. Scheduled tasks ## Inviting your team Once Appy.AI is installed in Slack or Teams, it's available to everyone in your organization. (Microsoft Teams lets you restrict access via app permissions if needed.) Add Appy.AI to a shared channel for team-wide use, or let team members find it in Apps to start their own direct message. # Meet your team Source: https://help.appy.ai/your-ai-team/meet-your-team A tour of the specialists on your AI team — and Violet, the Director of AI who orchestrates them. Screenshot 2026 05 20 At 3 17 44 PM Your AI team works the way a great team works: each specialist has a domain, they hand work off when it makes sense, and you don't manage the details. You talk to Violet — she handles the rest. New to this? See [Working with your team](/your-ai-team/working-with-your-team) for how to brief them, approve plans, and iterate on results. ## Meet Violet, your Director of Appy.AI Violet is the one you talk to. She holds a running picture of how your business operates — the tools, the rhythms, the open questions, what matters this week — and uses it to brief the right specialist for every job. Most AI assistants wait to be told what to do. Violet leads, suggests, and pushes back. She earns trust the way a great manager does: by knowing your business well enough that her recommendations land. ### What she does **Holds the picture.** Continuously builds a model of how your business operates — products, customers, tools, decisions in flight — so every brief lands with full context. **Routes the work.** Hears what you need, picks the right specialist, writes the brief, and brings back the result. You don't manage the handoff. **Asks the right question.** When she needs to clarify, it's one sharp question — not a survey. Knowing what to ask is most of the value. **Orchestrates the chain.** When a job needs Marcus to position, then Paige to write, then Maven to repurpose — Violet runs the sequence. You see the finished work. **Hires, mentors, and fires.** Spin up bespoke specialists when a job doesn't fit the existing team. Update an agent's playbook in plain language and the change takes effect immediately. Archive what you're not using. ## Your other AI team members Lincoln reads your site the way a crawler does — raw HTML, head signals, JSON-LD blocks, redirect chains, sitemap — and tells you the highest-leverage fix to make first. He doesn't ship the kind of SEO audit that's 80 pages of dashboard screenshots and zero recommendations. He ships the one paragraph that says *fix this, ship this, kill this — in that order.* Most SEO advice is noise. Lincoln cuts it. ### What he does * **Reads the site as the crawler sees it.** What Google actually indexes is the source of truth. Lincoln fetches the page, greps the head, checks the schema, traces the redirects — before he claims anything is broken. * **Builds keyword strategy grounded in real demand.** Finds the searches you should be ranking for that you're not. Cuts the vanity keywords. Maps intent to existing pages so the work is editing, not endlessly creating. * **Writes the schema markup.** JSON-LD for product, FAQ, article, organization, breadcrumb — the structured data that wins rich results and AI Overview citations. * **Maps content architecture.** Topic clusters, internal linking, page hierarchy — so authority compounds instead of getting scattered across 200 thin pages. * **Stays current on AI search.** Generative Engine Optimization, AI Overviews, LLM-driven discovery — the playbook is shifting, and Lincoln's advice anticipates where search is going, not where it's been. ### What he delivers * Technical SEO audit reports (PDF) with prioritized recommendations * Keyword strategy documents grounded in real search demand * On-page optimization plans, page by page * JSON-LD schema markup ready to drop into your site * Content architecture maps (pillars, clusters, internal linking) * Programmatic SEO blueprints ### Best with * **Site access** — Direct URL fetching (built in — no integration needed) * **Search analytics** — Google Search Console, Google Analytics (for query and traffic data) * **CMS / publishing** — WordPress, Webflow, HubSpot CMS (so recommendations can be picked up as tasks) * **Document storage** — Google Drive, OneDrive (for sharing the audit PDF) * **Delivery** — Slack, Microsoft Teams, email ### How it works Point Lincoln at a URL, brief him on the goal (audit, keyword strategy, schema, content architecture), and he produces a PDF you can read in ten minutes and act on this week. Recommendations come with priority and impact. Speculation gets cut. ### Try him with * "Audit our homepage for SEO and tell me what to fix first." * "Find 10 keywords we should be ranking for that we're not." * "Write the JSON-LD schema for our pricing page." * "Map a content cluster around \[topic] — pillar plus supporting pages." * "Is our site set up to win AI Overview citations? What's missing?" Marcus builds the strategic foundation your marketing, sales, and design teams execute against. Positioning, messaging architecture, voice and tone, audience profiles, naming — the decisions everyone downstream needs to know but nobody has time to make. He doesn't ship a workshop deck. He ships a Core Claim, defended, with ready-to-paste copy proving the strategy can land. Most positioning is crowded because nobody chose. Marcus chooses. ### What he does * **Owns the strategic answer.** Where to position. Who to serve. What to claim. The position no other competitor could honestly say. He picks one and defends it — instead of listing five and asking you to. * **Passes the Only test.** *"We are the only \[category] that \[differentiator]."* If three competitors could fill the same blank, the positioning isn't done. He sharpens until it is. * **Speaks in the audience's language.** Customers say *"my last vendor ghosted me,"* not *"service consistency challenges."* The words come from research, not internal decks. * **Anchors on structural advantage.** Proprietary methodology, audience specialization, integration depth, point of view — the things that compound. Cuts feature claims and "best customer support" theater. * **Ships strategy you can paste.** Every position recommendation comes with the homepage hero, the about-page opener, the sales-email subject line — proof the strategy survives contact with real copy. ### What he delivers * Positioning documents with the Core Claim, defended * Messaging architecture (audience, problem, promise, proof) * Voice and tone guidelines with examples * Audience profiles grounded in real customer language * Naming strategy (products, features, categories) * Marketing-psychology audits (what your current copy says vs. what it should) * Ready-to-paste hero, about, and sales copy ### Best with * **Research source material** — Google Drive, OneDrive, Notion (so he can read customer interviews, testimonials, and current site copy) * **Web access** — Direct site reads, competitor pages (built in) * **CRM context** — HubSpot, Salesforce (for win-loss notes and deal language) * **Delivery** — Slack, Microsoft Teams, email; outputs delivered as PDF or doc ### How it works Brief Marcus with what you have — current site copy, customer interviews, win-loss notes, competitor URLs — and he builds the position. He'll push back if your differentiator doesn't survive the Only test. The deliverable is always a decision, defended, with the first piece of paste-able copy attached. ### Try him with * "Position our product against \[competitor] in a one-pager." * "Build a messaging framework for our next launch." * "Audit our website copy for buyer-psychology gaps." * "Name our new \[feature/category] — give me three options with rationale." * "Rewrite our homepage hero — make it pass the Only test." Paige runs your content engine end-to-end — editorial pillars, calendar, long-form copy, visuals, repurposing, measurement. She treats content as a brand-as-publisher operation, not a traffic farm. One opinionated piece with the right visual outperforms five generic ones, and Paige knows which one to ship. She does the strategy and the craft. No handoff in between. ### What she does * **Leads with a point of view.** Generic content is the dominant failure mode. Paige finds the line worth taking and makes sure every piece reinforces it. * **Ships the system, not just the piece.** Editorial calendar, voice and tone, pillar/cluster map, repurposing flow, measurement loop. The piece is execution; the engine is the deliverable. * **Writes copy and conceives the visual together.** Most teams write first and brief a designer second — that's where voice dies. Paige treats the visual as part of the message. * **Builds for the channel.** A blog post and a LinkedIn carousel are different artifacts even with the same topic. Channel-native versions get planned in the brief, not retrofitted at the end. * **Applies editorial judgment, not just the prompt.** AI accelerates production. The thing AI can't do is reject the off-voice draft, the lazy pillar, the safe headline. That's where Paige's value lives. ### What she delivers * 90-day editorial calendars with pillars and cadence * Landing page copy (long-form, opinionated, channel-aware) * Blog posts and email sequences using proven frameworks * Content briefs (so even a junior writer can produce on-voice) * Hero images, blog illustrations, social-ready visuals * Pillar/cluster maps for SEO + content alignment * Repurposing plans (long-form → email → social → newsletter) * Content audits with kill/keep/revise calls ### Best with * **Document storage** — Google Drive, OneDrive (for source material and finished docs) * **Web access** — Direct site reads, competitor content (built in) * **CMS / publishing** — WordPress, Webflow, HubSpot CMS, Ghost * **Email platforms** — Mailchimp, Beehiiv, ConvertKit (for sequences) * **Image generation** — Built in (no setup required) * **Delivery** — Slack, Microsoft Teams, email; long-form delivered as Google Doc or PDF ### How it works Brief Paige on the business outcome (pipeline, brand, audience growth) and the channel, and she produces a system *and* the first artifact. You review, she iterates. The cadence becomes a rhythm — calendar set, pieces shipped, repurposing flow active. ### Try her with * "Build a 90-day editorial calendar for our blog." * "Write the landing page copy for our enterprise tier." * "Repurpose last month's webinar into four blog posts and a newsletter." * "Audit our blog — what should we kill, keep, revise?" * "Create a hero image for our new pricing page." Maven writes platform-native social content. One post per platform, on purpose. She knows the length budget of LinkedIn, the hook conventions of X, the first two seconds of a Reel, the unspoken rules of TikTok. Most social content is cross-posted from one platform onto five — Maven cuts that habit ruthlessly. She writes the platform first, then the post. ### What she does * **Writes platform-native, never cross-posts.** Same idea, different artifact. The audience can smell when a post wasn't written for them. * **Treats the hook as the work.** The first line on LinkedIn, the opening tweet of a thread, the first two seconds on a Reel — that's where disproportionate effort goes. Perfect body with a weak hook is a draft nobody reads. * **Conceives copy and visual together.** Voice dies on social faster than anywhere when the designer is briefed second. * **Reads performance data for judgment, not decoration.** When you upload last month's analytics, the deliverable is *what to cut next month* — not a slide deck of charts. * **Scripts the video.** 60-second Reel scripts, Short scripts, TikTok hooks, podcast-clip cuts — written for the platform they live on. ### What she delivers * Post packs by platform (LinkedIn, X, Instagram, TikTok, YouTube, Threads) * Social calendars with cadence and pillar logic * Video and Reel scripts with hooks, beats, and CTA * Ad copy with platform-specific variants * Social-shaped images (feed, story, carousel cover, quote card) * Social audits — what worked, what to cut, what the data is signaling * Storyboards for short-form video ### Best with * **Performance data** — LinkedIn Analytics, X Analytics, Instagram Insights, TikTok Analytics, YouTube Analytics (uploaded exports — Maven doesn't post on your behalf) * **Image generation** — Built in * **Document storage** — Google Drive, OneDrive (for asset libraries) * **Delivery** — Slack, Microsoft Teams, email; post packs delivered as spreadsheet, scripts as PDF ### How it works Tell Maven the campaign, the platforms, and the cadence, and she produces the post pack and the visuals. Upload last month's performance data and she returns a call about what to keep and what to kill. You publish manually — her job is making sure what you publish lands. Maven does not post, schedule, or connect directly to social platforms. ### Try her with * "Write a week of LinkedIn posts about our latest feature." * "Script a 60-second Reel for our customer story." * "Review this month's social data and tell me what's working." * "Build a 30-day social calendar for our product launch." * "Write three ad variants for \[campaign] — LinkedIn, Meta, X." Piper doesn't ship a list of 500 leads. She ships a list of 15 right ones, with the trigger that earns the read, the angle that earns the reply, and the email already drafted. Most outbound fails before the first send — wrong list, generic opener, three value props crammed into one paragraph. Piper fixes the upstream choices so the downstream activity actually compounds. The senior signal isn't emails sent. It's pipeline created. ### What she does * **Researches the prospect first.** ICP fit, decision-maker mapping, hiring and funding signals, recent news, tech stack. Finds the trigger that makes the email impossible to ignore. * **Writes ready-to-send cold outreach.** Single-email and multi-touch sequences. Relevance over personalization. One angle per email — committed. * **Builds follow-up sequences that earn replies.** Multi-touch cadence, breakup email at the right point. Stops at four. Honors the "no more emails" when someone asks. * **Rewrites for reply rate.** Hand her an underperforming sequence and she'll tell you why it's not working — usually wrong list before wrong copy. * **Generates LinkedIn connection requests and follow-ups.** The opener that's worth accepting; the bridge that's worth replying to. ### What she delivers * Targeted prospect lists (small and right, not large and generic) * Cold emails ready to send * Multi-touch sequences (3–4 emails, with breakup) * Follow-up and re-engagement emails * LinkedIn connection requests and DM sequences * Subject-line A/B variants * Email teardowns and rewrites with reasoning * Account research briefs ### Best with * **CRM** — HubSpot, Salesforce, Pipedrive, Close (for pulling and pushing leads) * **Sales engagement** — Instantly (for sequence sending) * **Lead enrichment** — Prospeo, LeadMagic (for finding emails and verifying contacts) * **Email** — Gmail, Outlook (so drafts land in your sent folder) * **LinkedIn** — Direct research and connection drafting * **Delivery** — Slack, Microsoft Teams, email ### How it works Give Piper a company, a target persona, or a current sequence to fix. She researches, finds the angle, drafts the email — and tells you the trigger that makes it work. You approve and send (or push to your sequencing tool). Pipeline starts compounding when the upstream choices are right. ### Try her with * "Write a cold email to \[company name] — find the right angle first." * "Build a 4-touch follow-up sequence for warm leads who went dark." * "Rewrite this email and give me three subject-line variants." * "Find me 15 right-fit prospects for \[product] and draft openers." * "Why isn't this sequence converting? Tear it down." Scout doesn't ship a deck of competitor profiles. He ships a recommendation: where to position, what to ship, who to beat, where to walk away. Every competitive move could be strategic or routine, and Scout is the one who tells the difference. The work compounds because he keeps watching. He won't fabricate funding numbers or company data. He flags where verification is needed. ### What he does * **Reads the field for signal.** Hiring, partnerships, beta features, pricing changes, leadership moves — sorts the meaningful from the cosmetic. * **Names the strategic question.** A competitive landscape isn't useful until someone says what it forces — where to position, what to ship, who to beat. Scout names it. * **Anchors on structural advantage.** Proprietary methodology, audience focus, integration depth, point of view — these compound. Feature gaps close in a sprint and don't. * **Builds tactical enablement.** Battle cards that live in the seller's pre-call prep. Comparison matrices sales can pull up in five seconds. Trackers that stay current. * **Calls when to walk.** Sometimes the right move is concede a wedge and own a different one. Scout will say so. ### What he delivers * One-page competitor write-ups (matrix + observations + recommendations) as PDF * Battle cards for sales pre-call prep * Feature and pricing comparison matrices * Positioning and messaging analysis (what they claim, where it's weak) * Win/loss analysis frameworks * Market landscape maps * Strategic recommendation memos * Competitive trackers that stay current over time ### Best with * **Web access** — Direct site reads, pricing pages, public docs (built in) * **LinkedIn** — Hiring signals, leadership moves, public sales motion * **Document storage** — Google Drive, OneDrive (for source material and outputs) * **CRM** — HubSpot, Salesforce, Pipedrive (for win-loss notes) * **Delivery** — Slack, Microsoft Teams, email; one-pagers delivered as PDF ### How it works Name the competitors, the question, and the audience for the deliverable. Scout researches, builds the matrix, and writes the recommendation. He'll flag any company data he couldn't verify so you don't quote a funding figure that's six months stale. ### Try him with * "Give me a one-pager on \[competitor] — what they do, how they sell, where they're weak." * "Build a battle card for our top three competitors." * "Compare our pricing to the market — where do we look strong and where do we look weak?" * "Map the \[category] landscape — who matters, who's noise, who to watch." * "What's \[competitor] been hiring for? What does it tell us?" Alexander works the deals already in motion. Inbound leads, booked meetings, open opportunities — he qualifies them hard, keeps the pipeline honest, and carries them through close, renewal, and expansion. He doesn't generate lists or write cold outreach. He takes a real deal and figures out what it needs to advance. Most pipeline reviews are status updates. Alexander's are strategy sessions. ### What he does * **Qualifies rigorously.** BANT, MEDDIC, or your framework of choice. Flags deals that don't belong in the pipeline and tells you why. * **Preps for every meeting.** Research, discovery questions, objection anticipation, agenda — so reps show up ready, not winging it. * **Drafts follow-up sequences and proposals.** Tailored to the deal stage, the stakeholder, and the conversation that just happened. Never auto-sends. * **Runs renewals and account plans before they come due.** Surfaces at-risk signals, identifies upsell and cross-sell angles, and prepares the QBR before it's on the calendar. * **Keeps the forecast honest.** Period-over-period pipeline analysis, deal velocity, close rate by stage — so leadership sees what's real, not what's optimistic. ### What he delivers * Deal qualification scorecards (BANT/MEDDIC) * Meeting prep briefs with discovery questions and objection maps * Follow-up email sequences and proposal drafts * Account plans and renewal briefs * QBR decks and forecast summaries ### Best with * **CRM** – HubSpot, Salesforce (pipeline data, deal stages, contact records) * **Email** – Gmail, Outlook (drafting follow-up sequences for review) * **Calendar** – Google Calendar (meeting prep, renewal scheduling) * **Delivery** – Slack, Microsoft Teams ### How it works Give Alexander a deal, a meeting, or a pipeline snapshot. He reviews what you have, asks what's missing, and tells you exactly what the deal needs to move. Drafts go into your review queue — nothing sends without your sign-off. Pricing is never invented. ### Try him with * "Qualify this inbound lead against MEDDIC and tell me if it's worth pursuing." * "Prep me for tomorrow's discovery call with \[company]." * "Draft a follow-up sequence for a deal that went quiet after the demo." * "Our \[customer] renewal is in 60 days — build me an account plan." * "Review this quarter's pipeline and tell me where the forecast is soft." Sage works at fractional-CFO altitude — not bookkeeper, not accountant. Cash forecasts, budgets, KPI snapshots, affordability scenarios. The model that matches how the business actually operates, and the call about which number the owner needs to act on this week. Small businesses can be profitable on paper and still die. Sage knows that — and operates accordingly. ### What he does * **Builds the 13-week cash forecast.** Cash flow beats P\&L for survival. Annual projections are too far away to help an owner decide whether to make payroll or take a deposit. The 13-week is the operational tool. * **Driver-based forecasting.** Models the actual factors — customer count, churn, pricing, headcount, cost per unit — not last year's numbers nudged up 10%. Test "what if we hire two reps" in seconds. * **Names variance.** When actuals deviate from budget, Sage names the driver behind the gap. *"Revenue is $8k below plan because two deals slipped from March to April"* — not *"Revenue down $8k."* * **Surfaces the metric you're not tracking.** Owners track what's easy. Sage surfaces the metric actually moving the business — DSO when collections slip, gross margin trend when pricing erodes, CAC payback when growth stalls. * **Matches horizon to decision.** A 12-month plan doesn't help with a \$15k marketing decision this month. A 13-week cash forecast does. ### What he delivers * 13-week rolling cash flow forecasts * Annual budgets (driver-based) * Runway analysis with named risks * KPI snapshots tailored to the business * Variance analysis with driver attribution * Affordability scenarios (can we hire? can we sign the lease? can we extend the contract?) * Monthly financial briefs * Financial health snapshots for board or lender conversations ### Best with * **Accounting** — QuickBooks Online, Xero, NetSuite, FreshBooks, Wave, Zoho Books * **Banking and spend** — Brex, Ramp, Mercury, Plaid-connected accounts * **Revenue and payments** — Stripe, Square, Shopify * **CRM and pipeline** — HubSpot, Salesforce, Pipedrive, Close (for forward revenue visibility) * **Spreadsheets** — Google Sheets, Excel via OneDrive (custom budgets and models) * **Payroll** — Rippling, Gusto, ADP (for loaded cost per FTE) * **Delivery** — Slack, Microsoft Teams, email ### How it works Connect the accounting and banking tools, brief Sage on the decision in front of you, and he builds the model. You ask "what if," he reruns. He grounds every recommendation in real numbers — and tells you which number to act on first. ### Try him with * "Build me a 13-week cash forecast from our connected accounting." * "What can we afford if we hire two engineers next quarter?" * "What's our runway and what's the biggest threat to it?" * "Why is gross margin sliding? Name the driver." * "Give me a monthly brief — what should I know about last month's numbers in three minutes?" Audrey keeps your books current — not a spreadsheet, not a service, an always-on bookkeeper on your team. She categorizes transactions, reconciles accounts, runs month-end close, and tells you what's worth your attention. The senior move isn't faster data entry. It's catching the duplicate, the misclassification, the new vendor nobody named — before they turn into year-end pain. Most bookkeepers ship clean transactions. Audrey ships clean transactions *plus* the call about what changed this month. ### What she does * **Keeps the books current.** Categorizes invoices, receipts, bank statements, and expense reports against your chart of accounts. Reconciles every account, every month, so the trail back to source is always intact. * **Catches what doesn't fit.** A \$4k charge with no project tag. A duplicate software fee. A new vendor that's never appeared before. Audrey points at the unusual instead of coding it and moving on. * **Runs month-end close.** P\&L, balance sheet, AR aging, and a one-paragraph "here's what's notable" — packaged so an owner can read it in three minutes. * **Chases the receivables you'd otherwise let slide.** Drafts follow-up emails for overdue invoices. Sends only when you approve. ### What she delivers * Categorized transaction registers with flagged exceptions * Monthly bank and credit-card reconciliations * Month-end close packages (P\&L, balance sheet, AR/AP aging) * Cash flow snapshots * AR follow-up drafts ready for your review * A short, plain-language "what happened this month" summary ### Best with * **Accounting** — QuickBooks Online, Xero, NetSuite, FreshBooks * **Banking and spend** — Brex, Ramp, Mercury * **Document intake** — Gmail, Google Drive, OneDrive (for invoices and receipts forwarded or uploaded) * **Delivery** — Slack, Microsoft Teams, email (so close packages and exception lists land where you already work) ### How it works You connect the tools you already use, brief Audrey on what you want, and approve the close before it goes out. She runs the categorization and reconciliation continuously, flags the items that need your judgment, and waits for your sign-off on anything that goes to a customer or vendor. You're always the approver. She's the one with the books open. ### Try her with * "Categorize this month's bank statement and flag anything weird." * "Reconcile our credit card and tell me what doesn't match." * "Pull together a month-end close package." * "Show me which customers are over 60 days late and draft follow-ups." * "Walk me through what's notable about last month's books in three sentences." Sloane reviews every pay cycle before it posts, books the journal entry after it clears, reconciles against your bank, and produces the labor cost reports nobody else writes. The senior move isn't faster submission. It's catching the missing timecard, the unusual delta from last cycle, the contractor about to cross the 1099 threshold — *before* the cycle goes out. Calculating pay, withholding tax, issuing W-2s — that's your payroll provider's job, and it's regulated. The pre-cycle review, the JE, the reconciliation, the labor cost story — that's Sloane's. ### What she does * **Pre-cycle review.** Reads the upcoming pay register from your payroll provider and flags anomalies: missing time, unusual deltas vs. last cycle, new hires not yet set up, contractors crossing thresholds. Before you click submit. * **Books the journal entry.** After each cycle clears, books the payroll JE — gross wages, withholdings, employer taxes, net cash out — into your accounting tool so Audrey's reconciliation just works. * **Reconciles the bank withdrawal.** Matches the payroll provider's net cash out to the actual debit from your operating account. Surfaces anything that doesn't tie. * **Tracks 1099s as you go.** Watches every contractor payment so year-end 1099 prep is mechanical instead of a December scramble. * **Makes labor cost legible.** Labor cost as a percent of revenue, headcount by department, year-over-year deltas — payroll is usually the biggest line item, so it gets its own readable story. ### What she delivers * Pre-cycle anomaly reports with named flags ready for your decision * Payroll journal entries posted to your accounting tool * Bank reconciliations against each cycle's net cash * Labor cost reports (by department, by period, year-over-year) * Contractor 1099 tracking and year-end prep * Headcount snapshots ### Best with * **Payroll** — Gusto, ADP, Rippling, Paychex, OnPay * **Accounting** — QuickBooks Online, Xero, NetSuite, FreshBooks (for booking the JE) * **Banking and spend** — Brex, Ramp, Mercury (for reconciliation) * **Time tracking** — Harvest, Toggl, Clockify (when separate from the payroll provider) * **Delivery** — Slack, Microsoft Teams, email (so pre-cycle reports and labor cost summaries arrive where the cycle gets approved) ### How it works You connect the tools you already use, Sloane reads the cycle and flags what's off, you approve. After the cycle posts, she books the JE, reconciles the bank, and updates labor cost reporting. Every cycle is approved by you — payroll going out wrong is harder to fix than payroll going out late, and Sloane is built around that fact. ### Try her with * "Review this pay cycle before I submit — flag anomalies." * "Book the payroll journal entry and reconcile against the bank." * "Show me labor cost by department for the last quarter." * "Which contractors are getting close to the 1099 threshold this year?" * "Compare this cycle to last and tell me what changed." Sarah synthesizes across email, calendar, Slack, notes, and docs into one output. She briefs you on the day, triages your inbox, drafts in your voice, and turns meeting transcripts into action items. The senior signal isn't activity volume — it's you making decisions faster, staying informed without overload, and trusting that nothing important falls through the cracks. She leads with what needs action. Context comes after. ### What she does * **Briefs the day.** What's on your calendar, what landed overnight, what's at risk of slipping. One coherent read — not a tool-by-tool dump. * **Triages the inbox.** What actually needs you, what can wait, what you can ignore. Drafts replies in your voice for the things you'd reply to anyway. * **Runs the calendar.** Schedules, manages density and buffers, flags conflicts before they happen, gatekeeps requests that aren't worth the time. * **Turns meetings into action.** Transcripts become decisions, action items, and decision logs. Multiple sessions get rolled up. Owners and dates attached. * **Drafts in your voice.** Professional but human. No filler, no corporate padding. Offers to adjust after presenting — doesn't ask for context she can reasonably infer. ### What she delivers * Daily briefings (calendar + inbox + signals from connected tools) * Inbox triage with named tiers and draft replies * Meeting summaries with action items and decision logs * Calendar audits and buffer recommendations * Drafted messages in your voice (email, Slack, Teams) * Rolled-up summaries across multiple sessions * Gatekeeping recommendations (what to decline, defer, delegate) ### Best with * **Email** — Gmail, Outlook * **Calendar** — Google Calendar, Microsoft Calendar * **Messaging** — Slack, Microsoft Teams * **Docs** — Google Docs, OneDrive, SharePoint * **Notes** — Notion (for project context and saved preferences) * **Meeting transcripts** — Uploaded recordings, transcripts, or pasted notes * **Delivery** — Slack, Microsoft Teams, email ### How it works The more tools Sarah is plugged into, the more she can synthesize. Brief her on the day, the meeting, or the message — and she comes back with one coherent output. Drafts wait for your approval before they send. The senior move is knowing what *not* to surface, schedule, or draft. ### Try her with * "Give me a briefing on what I'm walking into today." * "Triage my inbox — what actually needs me?" * "Turn this meeting transcript into action items and a decision log." * "Draft a reply to \[name] about \[topic] in my voice." * "Audit my calendar this week — where's the density problem?" Dexter monitors operational data and surfaces what's off — by how much, and whether it's recurring. He doesn't ship a wall of numbers. He ships the variance that matters, the pattern behind it, and a clear note about what needs attention. Staff get a plain action item. Managers get trend context. Leadership gets the strategic pattern and the items that need a decision. Same data, different rendering. ### What he does * **Leads with what's off.** The variance, the flag, or the anomaly comes first. Period context comes after. A reader knows within ten seconds whether they need to act. * **Separates one-off from pattern.** A single variance is a flag. The same variance three weeks running is a problem. Always notes which. * **Matches depth to audience.** Staff need a specific next step in plain language. Managers need trend plus recommended action. Leadership needs the strategic pattern. * **Flags data problems directly.** Quietly normalizing bad data is how trust erodes. Dexter says so when a feed is incomplete or a category is inconsistent. * **Anchors every claim in specific numbers.** Expected $4,200, actual $3,440, delta –\$760. Vague numbers waste the reader's time. ### What he delivers * Variance reports (with named drivers, not just gaps) * Period-over-period comparisons (week-over-week, month-over-month, year-over-year) * Anomaly detection summaries (one-off vs. recurring) * Operational dashboards translated into prose * Audience-tailored briefs (staff / manager / leadership) * Data quality flags (what's missing, what's inconsistent) * Recommended-next-step memos ### Best with * **Spreadsheets** — Google Sheets, Excel via OneDrive (for operational data) * **Accounting** — QuickBooks Online, Xero (for financial-adjacent ops data) * **POS and inventory** — Square, Shopify, Toast (for retail and food service) * **Project management** — Asana, ClickUp, Linear (for throughput and cycle data) * **CRM** — HubSpot, Salesforce, Pipedrive (for sales-ops metrics) * **Document storage** — Google Drive, OneDrive (for source files and outputs) * **Delivery** — Slack, Microsoft Teams, email ### How it works Point Dexter at the data — connected tools or uploaded exports — and tell him the audience. He produces the right shape: a one-liner for the staff member, a trend chart for the manager, a strategic memo for the owner. You don't dig. ### Try him with * "Compare last month's ops numbers to the month before — flag what changed." * "Spot the anomalies in this quarter's data and tell me which ones repeat." * "Build a leadership-ready summary of our ops performance." * "Inventory is down — is it a blip or a pattern?" * "Audit our supplier on-time delivery for the last 30 days." Joy treats every ticket as signal — a churn flag, a product gap, an expansion opening. She doesn't get to inbox zero. She tells you what the customers are actually trying to accomplish, ranks the signals by severity and pattern, and drafts the communication that matches. Apology theater wastes everyone's time. Joy acknowledges once and leads with what's being done. ### What she does * **Triages by intent and severity.** What's the customer actually trying to accomplish — not just what they're asking for? What's the impact, urgency, sentiment? Severity-matched. * **Reads patterns across tickets.** Five customers reporting the same issue is one insight with volume, not five separate items. Consolidates ruthlessly. * **Quotes customers verbatim.** Their actual words preserve nuance. One or two sentences of quote beats a paragraph of summary. * **Drafts severity-matched responses.** Calm acknowledgment for low-impact. Direct ownership for critical. Tone is part of the work. * **Surfaces retention and expansion signals.** Tier-aware: an annual-renewal customer with escalating language gets flagged differently than a new free-tier user with the same complaint. ### What she delivers * Triaged ticket queues with severity, intent, and recommended action * Customer communication drafts (severity-matched, ownership-driven) * Insight reports (what patterns are emerging across volume) * Sentiment summaries (with verbatim customer quotes) * Bug-vs-feature-request classification * Escalation briefs for product or engineering * Review monitoring summaries (App Store, G2, Trustpilot, Google) ### Best with * **Support platforms** — Zendesk, Intercom, Freshdesk, Help Scout (for ticket pulls) * **Email** — Gmail, Outlook (for ticket sources and reply drafting) * **Reviews** — Uploaded exports from G2, Trustpilot, App Store, Google Reviews * **CRM** — HubSpot, Salesforce, Pipedrive (for customer tier context) * **Communication** — Slack, Microsoft Teams (for routing escalations and posting summaries) * **Document storage** — Google Drive, OneDrive (for transcripts, screenshots, CSVs) * **Delivery** — Slack, Microsoft Teams, email ### How it works Hand Joy the tickets — connected platform or pasted/uploaded batch — and tell her the audience. She returns triage with severity, the patterns nobody's named yet, and draft responses ready for your review. The fix isn't "more replies" — it's the systemic call you can defend. ### Try her with * "Triage this week's support tickets — what needs me, what can wait?" * "Read these 50 reviews and tell me what customers actually feel." * "Draft a response to this escalated complaint." * "What's the pattern across this month's tickets?" * "Five customers are complaining about \[issue]. What do I do?" Cole generates and edits actual video assets. Give him a brief, a script, or raw footage — he builds the finished clip. Short-form social content, product demos, animated explainers, ad creative, motion graphics — Cole handles the full production job, not just the generation step. Most AI video tools stop at the clip. Cole finishes the job. ### What he does * **Generates video from a prompt or still image.** Short clips created from text descriptions or animated from an existing image — useful for social content, explainers, and ad creative when you don't have footage. * **Edits and assembles footage.** Trim, splice, cut to rhythm, layer B-roll — takes raw or generated clips and builds a finished piece. * **Formats for platform.** Aspect ratio conversions (16:9, 9:16, 1:1), duration cuts per platform spec, compression for upload. * **Adds captions and subtitles.** Auto-generated and timed to the audio track. * **Sourced music and overlays.** Background audio, lower thirds, motion graphic elements where the brief calls for them. ### What he delivers * Generated video clips (from prompt or still image) * Edited and assembled video files (MP4) * Platform-formatted versions (Reels, TikTok, YouTube, LinkedIn) * Captioned/subtitled cuts * Product demos and animated explainers ### Best with * **Uploaded footage** – provide raw clips directly * **Script or brief** – the tighter the brief, the better the output * **Delivery** – Slack, email (for sharing finished files) ### How it works Brief Cole on what you're making — platform, length, tone, any footage you have — and he takes it from there. For generation, a text prompt or still image is enough. For editing, upload your footage and describe the cut. Output is a finished video file ready to publish. ### Try him with * "Create a 15-second product demo clip for Instagram Reels." * "Animate this product screenshot into a short explainer video." * "Edit this raw interview footage down to a 60-second highlight cut." * "Add captions to this video and reformat it for TikTok (9:16)." * "Make a short ad video for our new feature launch." Hamilton builds complete static websites end to end — brief in, live URL out. He owns the whole job: HTML structure, CSS system, copy when the brief is thin, images when the design calls for them, favicon, OG image, meta tags, and deploy. You don't manage a handoff between a designer and a developer. You just get a live site. Most website tools produce templates. Hamilton produces sites. ### What he does * **Designs and builds from scratch.** Semantic HTML, separate CSS files, mobile-first responsive layouts, modern CSS (grid, flexbox, custom properties). No page builders, no bloat. * **Handles copy when needed.** If the brief is thin, he writes the headlines, body copy, and CTAs — then you refine. * **Sources and integrates imagery.** Stock photography, AI-generated visuals, uploaded assets — placed and styled to the design system. * **Takes care of the details.** Favicon set, Open Graph image, meta title and description, proper heading hierarchy, and load performance. * **Deploys to a live URL.** Publishes the finished site to a public link you can share immediately. * **Pulls in libraries when they earn it.** Chart.js for data, Alpine.js for interactivity, HTMX for dynamic behavior — CDN-loaded, only when the design calls for them. ### What he delivers * Complete static websites (HTML/CSS) * Landing pages, microsites, one-pagers * Event and RSVP pages, portfolio sites, marketing pages * Lead-gen pages, product pages, link-in-bio pages * Live public URL on completion ### Best with * **Brand assets** – logos, color palette, fonts (speeds up the design pass) * **Copy** – headlines, body text (use what you have; Hamilton fills the gaps) * **Delivery** – Slack, email (for sharing the live link) ### How it works Give Hamilton the brief — type of site, purpose, audience, any assets you have. He designs, builds, and deploys. The live URL comes back when it's done. Changes go back to Hamilton as follow-up requests. ### Try him with * "Build a landing page for our new product launch — here's the brief." * "Create a one-page event site with an RSVP form for our webinar." * "Make a simple portfolio site for our agency." * "Build a lead-gen page for our free trial offer." * "I need a link-in-bio page for our Instagram — clean, minimal, on-brand." Margot builds brand-consistent visual assets end to end. She starts by extracting a brand profile — from a logo, website, or brand guide — and uses it to lock palette, typography, lighting, and composition across every asset she creates. Single pieces or coherent packs, generated from scratch or edited from what you have. Most AI image tools produce output. Margot produces assets that fit your brand. ### What she does * **Extracts and maintains a brand profile.** Pulls the visual system from whatever source material you have — logo, site, existing deck — and holds it consistent across every asset in the project. * **Generates brand-aligned visuals.** Hero images, marketing graphics, social headers, posters, illustrations, icons — designed to the extracted brand, not generic stock. * **Produces logo and identity assets.** Logo variations, favicon, OG image — the foundational pieces a brand needs. * **Edits existing images mechanically.** Resize, format conversion, compression, watermarking, print prep (DPI, bleed, CMYK guidance) — deterministic operations on files you already have. * **Builds asset packs.** Coherent sets for a campaign, a product launch, or a social channel — consistent across every piece. ### What she delivers * Brand profile (extracted palette, typography, style rules) * Logo variations, favicon, OG image * Hero images and marketing graphics * Social headers, feed graphics, poster templates * Icon sets and illustrations * Resized, converted, compressed, or watermarked image files ### Best with * **Existing brand assets** – logo files, brand guide, website URL (for brand extraction) * **Reference images** – examples of the style you want * **Delivery** – Slack, email, Google Drive, OneDrive ### How it works Give Margot a brief and any brand assets you have. She extracts your visual system first (or confirms what you've described), then produces assets to that spec. For mechanical edits, upload the file and describe the operation. She flags where source material needs to come from you rather than inventing it. ### Try her with * "Create a hero image for our product launch page — here's our brand guide." * "Build a social header pack for LinkedIn and Twitter." * "Extract our brand palette from our website and produce an icon set." * "Resize these 10 images for web and compress them under 200KB." * "Design a favicon and OG image that match our logo." \ Ready to put them to work? See [Working with your team](/your-ai-team/working-with-your-team). # Overview Source: https://help.appy.ai/your-ai-team/overview Staff up with an entire team of AI coworkers Appy.AI gives you a complete AI team from day one — specialized agents that work around the clock to scale your business, right where your team already operates. No hiring, no training, no setup. Org Chart ## Meet Violet — your Director of Appy.ai Every team needs a leader. Violet is yours. She coordinates your entire AI team — translating your goals into plans, then delegating work to the right specialists automatically. Working with Violet: 1. **Share your business** — Tell Violet what you're working on and what matters most. 2. **Approve the plan** — Review her strategy, give feedback, then watch her delegate. 3. **Track progress** — See your team take action, or kick off new tasks in parallel. ## What you get out of the box Your AI team comes pre-built with over a dozen specialists, each ready to tackle a different part of your business. * **Marketing** — Positioning, content, SEO, and platform-native social * **Sales** — Outbound prospecting and competitive intel * **Finance** — Cash forecasting, bookkeeping, and payroll review * **Operations** — Daily briefings, ops analytics, and customer service triage Ask Violet for your full roster at any time. Need a specialist that doesn't exist yet? Just describe the role and Violet will hire them. **New here?** Start with [Meet your team](/your-ai-team/meet-your-team) — a tour of the specialists and how Violet orchestrates them. ## How your team works Working with your AI team feels like messaging a colleague. Open Slack or Microsoft Teams, describe what you need in plain language, and Violet routes the work to the right agent. When a task needs multiple specialists — a blog post that needs writing, SEO, and a header image, for example — they coordinate behind the scenes. Results come back to you in the same thread where you asked. You can also set up agents to run on schedules, handling recurring work around the clock without you having to ask. ## What your agents actually produce Your agents deliver real work — not just chat responses: * Blog posts, emails, and social content * Competitor research and summaries * Meeting summaries with action items * Images for content and campaigns * Sales outreach emails * SEO-optimized rewrites * Presentations, spreadsheets, and formatted reports # Working with Your Team Source: https://help.appy.ai/your-ai-team/working-with-your-team Define goals, approve plans, and track results — your agents handle the rest. ## Assign tasks Your agents respond to plain language — describe what you need the way you'd ask a coworker, and Violet routes it to the right specialist. In Slack, open a direct message with Appy.AI. In Teams, send a DM or @mention an agent in a channel — they'll respond in the thread. Here are some examples: * "Draft a follow-up email for the client meeting yesterday" * "What are the SEO opportunities for our homepage?" * "Research what our top 3 competitors are doing on social media" * "Summarize the team standup recording and pull out action items" * "Create a header image for our blog post about spring promotions" * "Write a cold outreach email for the leads list I uploaded" Need a proposal or report? Tell Violet what you need — they'll deliver a formatted Word document, complete with tables, images, and charts, ready to review or send. **Haven't met the team yet?** Start with [Meet your team](/your-ai-team/meet-your-team) for a tour of all specialists. ## How agents work together Some tasks need more than one specialist. Violet coordinates them automatically — you just describe the outcome you want. Say you ask for a blog post about your new product launch. Paige drafts the article, Lincoln optimizes it for search, and Margot generates a header image. You get the finished result — not three separate handoffs. Or you're preparing for a client pitch: Scout pulls market intel, Marcus aligns the messaging, and Paige drafts the outline. One cohesive deliverable, back in your thread. ## Autonomous work Your agents can run around the clock — no prompting required. Set up schedules and triggers to handle recurring tasks automatically. Some examples: * Weekly competitive intelligence digests * Daily summaries of industry news * Automatic meeting summaries after every call * Regular content performance reports To set it up, just tell Violet: "Send me a competitive update every Monday morning." ## Starting a new topic A long conversation drifts — old context piles up and can shape answers you don't want it to. Type **New Topic** in any Slack thread, Teams chat, or Telegram chat, and your agent ends the current conversation and starts a fresh one right there, with no earlier context carried forward. This is the fastest way to reset without leaving the thread or losing your place: * **Mid-task, it still finishes the work.** If your agent is in the middle of something when you send the command, it wraps up and delivers what it was doing first, then starts the new session. * **Idle, it resets immediately.** If your agent is just waiting on you, the new session starts right away. Case and extra spacing don't matter — "new topic," "New Topic!", and " NEW TOPIC " all work. You can also just ask: "How do I start over?" and Violet will tell you to type **New Topic**. ## Checking in on results Results come back in Slack or Teams, right in the thread where you made the request. Every instruction, response, and decision is visible in the activity stream. If an agent runs into an issue, it'll tell you in the chat — no need to check the web dashboard. Want changes? Just say so: * "Make it more concise" * "Add a section about pricing" * "Can you make the tone more casual?" Your agents refine based on feedback. Keep iterating until it's right. The web platform keeps a full history of conversations, completed work, and audit logs — every task has a clear owner, status, and history.