Source: https://actvox.com/en/help/user-guide # ActVox user guide ActVox is a **shared workspace that people write in and AI agents connect to**. It combines your company context, working pages, databases, research, content, conversations, and agent actions in one workspace. This guide explains the current product experience for day-to-day users. Features can be permission-, workspace-, and integration-dependent. ActVox does not publish externally or perform destructive work without the relevant approval path. ## In this guide 1. [Get started](#1-get-started) 2. [Workspace OS](#2-workspace-os) 3. [Create and format pages](#3-create-and-format-pages) 4. [Build databases](#4-build-databases) 5. [Share and collaborate](#5-share-and-collaborate) 6. [Work with the agent](#6-work-with-the-agent) 7. [Automation and connections](#7-automation-and-connections) 8. [Admin and security](#8-admin-and-security) 9. [Reports](#9-reports) 10. [Strategy](#10-strategy) 11. [Design.md](#11-designmd) 12. [Knowledge base](#12-knowledge-base) 13. [Drive](#13-uploaded) 14. [Memory](#14-memory) 15. [Conversations](#15-conversations) 16. [Studio](#16-content) 17. [Artifacts](#17-artifacts) --- ## 1. Get started ### ActVox basics A workspace is the unit where ActVox keeps your company context and work. Use one workspace for one brand, client, or operating team so the agent can reason from the right business context. Your workspace can contain: | Object | Use it for | | --- | --- | | **Pages** | Plans, briefs, notes, meeting records, operating documents, and structured evidence. | | **Databases** | Reusable tables, boards, calendars, and lists for campaigns, tasks, content pipelines, competitors, or research. | | **Context** | Your business profile, standing instructions, strategy, reports, sources, uploads, and memory. | | **Conversations** | Work with the agent while preserving the workspace context. | | **Outputs** | Content and agent-generated artifacts that can be reviewed and reused. | ### First workspace setup 1. Sign in. A new workspace opens on two doors: **Connect my AI** or **Start blank**. 2. **Connect my AI** copies a prompt into Claude, Codex, ChatGPT, or another client. Read and write are the powers it asks for. What it writes shows up here as pages you can open. 3. **Start blank** opens Home. Home has a short guide and examples. Remove the examples when you want a clean page. 4. Home then shows **Get started**, which says what to do next. When it reads **You're set up**, that heading opens **Start here**. 5. Add source material in [Knowledge base](#12-knowledge-base). Files in [Drive](#13-uploaded) are not searched until you add them there. 6. Open **ACTVOX.md** for standing instructions the agent should consider on every turn: approval boundaries, terminology, voice, target geography, or source-quality rules. A teammate invited into an existing workspace still confirms the company profile before the workspace opens. Company website, description, audience, and products stay editable in **Settings**. ### Home Home opens with a greeting and a **Book a demo** button. The button books a short call with the founder. Home has no ask box. To talk to the agent, press ⌘T to open a chat tab, or open a chat from the sidebar. Under the greeting, Home shows example sections with real blocks and four guide sections. **Bring your context from Claude, ChatGPT, DeepSeek or any AI** has one toggle for each assistant. Each toggle gives the steps and a prompt to copy. Claude and ChatGPT write the pages here through their connector. DeepSeek and other assistants give you Markdown: paste it into a page, or save it as a .md file and upload it to **Knowledge Base**. **Build a dashboard** shows number tiles that read the sample database live. **Interactive HTML: how it works** tells what an HTML block can do: it runs in a closed frame, reads one bound database live, uses ECharts and three.js, and exports to video. An agent that builds a dashboard follows the dashboard rules in the `actvox-html-artifact` skill (`references/dashboard-rules.md`): small steps, clear states, a check at phone and desktop widths, and size limits. **Videos and screencasts of any app** shows the Motion block and animated HTML. Both export to MP4 with **Export video**. To record a screencast, press ⌘ Shift 5 on a Mac or open Snipping Tool on Windows. Then drop the file on the page. Every prompt is a code block with a **Copy code** button. Two more **Book a demo** buttons sit in the guide. **Remove the examples** on Home removes the guide too. **Undo** puts it back. ### Start here **Start here** is a normal page under Home. It says how the workspace fits together, in five blocks: **Connect your AI**, **Organize your workspace**, **Your sidebar**, **Knowledge Base and Memory**, and **Studio**. Each block is about five lines and one **Try it**, which opens the real place. Notes you type between the blocks stay on the page. The title follows your language until you rename it. The page is not shared. Open it from the account menu (**Learn ActVox** › **Open the Start here guide**) or from the command palette. If it was in Trash, that opens the same page again. **What a page can hold** is a page under Start here that shows the rich blocks working: a sample board and the same table as a linked view, a chart bound live to them (move a card and the bars change), columns, a grid, a callout, a toggle, a photo gallery, a 3D model and a Motion block. A one-minute tour sits at the top: an HTML block that plays real screenshots. Press play. Its words are in the block's data, so your agent can change them. Each section says how to insert the block (the `/` command) and gives a prompt to send to your agent. Every block is real: edit it, copy it to your own pages, or delete it. The sample page and its rows do not count against your block allowance. Open it from the account menu (**Learn ActVox** › **See what a page can hold**), the command palette, **Get started** on Home, or **See every block** on Start here. The first time you open Memory, Drive, Reports, Content, Calendar, or Brand DNA, a one-line strip explains the area, offers its main action, and links **How it works** to the matching Start here section. Knowledge Base, Artifacts, and pages get a short tour instead, pointing at the real controls. Each shows once. ActVox starts at most one of these, or any other tip, on its own every six hours, and none in your first day or while **Get started** is on Home. **Take the tour** under **Learn ActVox** in the account menu walks the screen you are on whenever you ask for it. In workspaces where ActVox offers the **guided tour**, it starts once on your first visit after you choose how to start, or when you join a team from an invite. It walks real steps, not slides: the sidebar shows one row at a time, and a step that asks you to do something (move a card, assign a task to your agent, accept its work) moves on when you do it. A step you only read has **Next**, and **Skip this step** and **Skip tour** are always there. Halfway, a checkpoint shows your **Get started** progress and lets you stop. To run it again, open **Take the tour** in the account menu (resume, start over, or the tour of this screen) or the command palette. It resumes where you stopped, on any device. > **Good operating pattern:** establish the business profile and source material before asking for a consequential strategy, report, or content plan. Better context produces more useful and more attributable output. ### Sidebar navigation The left sidebar is the map. | Row | What you will find | | --- | --- | | **New** | A chat, a page, a database, or a whiteboard. | | **Home** | The workspace home page, and the pages filed under it, including Start here. | | **Page tree** | Pages and databases nested together. The row is labelled with the workspace name. Drag one page onto another, and pin what you use most. | | **Tools** | Chats, Reports, Content, Knowledge Base, Drive, and Brand DNA. Only **Drive** is under **Your tools** until you choose otherwise; the rest wait under **More**, and the sliders next to **Your tools** choose which ones show. **Billing** shows here only when your usage runs low. | | **Pinned** | Your private shortcuts. | Some tools hold more than one view. **Reports** has Reports and Strategy. **Content** has Cards, Gallery, and Calendar. **Knowledge Base** has Library and Memory. **Drive** has Artifacts in its left rail. Activity is the **Workspace** view of **Inbox**, and Billing is in **Settings**. Content and Brand DNA are the Studio tools: drafts, when each draft goes out, and the colours, type, logo, and voice every image and draft follows. **Skills** live in **Squad**: choose **Skills** on its filter row, or **Manage skills** in the chat composer’s ✨ menu. **Agents & API** (called Connectors until November 2026) is the one place that connects agents: open it from the account menu (your avatar at the foot of the sidebar), from **Connect your AI**, or from **Invite Agent** in Squad. --- ### Get the app ActVox runs in your browser and installs as an app on your phone or computer. The steps for your device are on [the Get the app page](https://actvox.com/en/apps). In the app, **Settings → Apps** shows the same install card, and a QR code opens it on your phone. The Mac download is offered there too. The desktop app for Windows and Linux is on the waitlist on that page. ## 2. Workspace OS The Workspace OS is the main work surface. It keeps live documents, structured work, context, and agent output connected to the same workspace. **Get started** on Home says what to do next. **Start here**, a page under Home, says how the pieces fit. Open it from the account menu or the command palette. ### Workspace settings Open **Settings** from the workspace controls. **Settings → Workspace** keeps the workspace identity and marketing profile. Each member sees the result. | Setting | Why it matters | | --- | --- | | Workspace/project name | Identifies the workspace in navigation and switching. | | Business name and logo | Gives the workspace a clear company identity. | | Website URL | A starting point for company analysis and source context. Use a complete `https://` or `http://` URL. | | Location and industry | Helps ground local, market, and audience work. | | Business description | A concise explanation of what the company does. | | Brand voice and target audience | Improves content, messaging, and strategic recommendations. | | Key products and unique selling points | Gives the agent durable commercial context. | | Brand colors | Supports brand-aware work and media direction. | | **Language & theme** | Your interface language and theme are in **Settings → Account**: one default for every workspace. **Settings → Workspace** sets a choice for this workspace when you want one. | Click **Save** after changing settings. **Delete Project** is in **Settings → Advanced**, under **Danger Zone**, last. To confirm, you type the workspace name. ### Settings and preferences **Settings** has six tabs. It opens on **Account**. What you see can depend on your role, your plan and the deployment. | Tab | Use it for | | --- | --- | | **Account** | Your name, email addresses, sign-in and devices, your default language and theme, the agent reply language, the app, your cookie choice and the legal documents. **Manage account** in the account menu opens it. Only you see these settings. | | **Workspace** | **General** (name, link name, business details), **Access**, **Auto-adapt**, **Customize**, the language and theme of this workspace, and **Shared links**. Each member sees the result. | | **Organization** | The name and logo of the organization, and who can create workspaces. The gear on an organization row opens it. You manage people in **Squad**. | | **Billing** | Your plan, usage, invoices and billing contact. See **Billing** below. | | **Notifications** | What reaches you. Only you see these settings. | | **Advanced** | **Engineering** for organization admins, **Export workspace data**, and **Danger Zone** with **Delete Project**, last. | The settings of the agent are on the profile of the built-in agent: **Squad** → **ActVox** → **Settings**. Answers follow your reply language (**Settings → Account**). If you did not choose one, they follow your interface language. A workspace default applies only to shared work and to people who chose nothing. #### Billing **Settings → Billing** is the one page for your plan. **Manage billing** in the account menu opens it. The **Billing** row in the sidebar opens it too, when the row shows. - Four facts start the page: **Plan** with its price each month, **Renews** (or **Ends**), **Usage** as the percent used this period, and **Payment method**, the card on file, with **Update**. - **Invoices** lists each invoice with its number, date, amount and status. **Pay now** opens a due invoice on the Stripe page, where you pay it. **Download** gets its PDF. Older invoices are in the billing portal. - **Billing contact** shows the address for invoices and receipts. An organization also names its accounting person. - In an organization, only an admin sees the invoices and changes the billing contact. #### Agent preferences On the built-in agent's profile, **Squad** → **ActVox** → **Settings** (or **⋯** → **Agent settings**), you can set the agent name, default response language, greeting, and communication style. Available built-in personas include Professional, Friendly, Concise, Detailed, and Custom. For standing instructions, use **ACTVOX.md** in the Context section of the workspace tree. It is the current editable instruction surface that the agent reads on every turn. The older system-instructions settings field is retained for legacy setup but is not the place to maintain live instructions. ### Workspace navigation Use tabs to work across multiple pages, databases, conversations, and outputs without losing your place. The active tab appears in the center workspace area. Open related work in separate tabs when comparing a brief to a database, report, or agent result. Useful navigation habits: - Keep business-wide guidance in **ACTVOX.md**, **Profile**, and [Strategy](#10-strategy), not scattered across campaign pages. - Keep the working system - planning, execution, and tracking - under **Workspace**. - Put raw evidence in [Knowledge base](#12-knowledge-base) or [Drive](#13-uploaded), then cite or link it from pages. - Use [Content](#16-content) and [Artifacts](#17-artifacts) as output review queues rather than as the only source of truth. --- ## 3. Create and format pages Pages are durable documents for briefs, research notes, plans, reports, playbooks, and decisions. ### Create a page You can create pages from the Workspace tree or from inside another page. 1. In **Workspace**, use the add control for the target location. 2. Choose **Page** and give it a clear, action-oriented title. 3. To create a nested page, open the parent page and insert a **Page** block, or create it in the corresponding subtree. 4. Start writing. Changes save automatically after a short pause. A useful naming convention is: `YYYY-MM - outcome - audience or channel`, for example `2026-08 - LinkedIn founder narrative - B2B SaaS`. ### Use blocks and slash commands Type `/` in the page editor to open the block menu. Search by name, then select a block. You can also use the block handle to insert, move, or change a block. | Block group | Examples | | --- | --- | | **Basic** | Text, headings 1-4, bulleted and numbered lists, to-do lists, toggles, callouts, quotes, simple tables, dividers, child pages, and page links. | | **Media** | Images, video, audio, files, web bookmarks, and code blocks. | | **Database** | Inline database, full-page database, linked database view, and table, board, gallery, list, or calendar views. | | **Advanced** | Table of contents, equations, breadcrumbs, and two- to five-column layouts. | | **Embeds** | URL embeds and sandboxed HTML where enabled. | ### Format and organize pages - Use headings for a scannable hierarchy; table of contents blocks use that hierarchy. - Use a callout for the decision, assumption, risk, or owner that a reader must not miss. - Use to-do blocks only for lightweight page-local work. Use a database when work needs owners, dates, reporting, or repeatability. - Add a page icon or cover to distinguish high-traffic documents. - Link related pages rather than copying their content. This keeps evidence and decisions current. - Use the comments panel for collaboration and the version/history controls for review when available. - On a phone, you can comment while staying in Reader. Select text and choose **Add comment** to discuss that passage. Editing permission is required. - When an agent suggests changes on a phone, choose **Review** to enter Edit and jump to the first change. **Accept all** applies the pending changes; **Reject all** dismisses the suggestion. Editing permission is required. - To change block order on a phone, enter **Edit** and choose **Reorder**. Use the up/down buttons to move a block; each move has its own undo step. Close the sheet or use Back to return to the page. Reordering is unavailable while an agent suggestion is pending. ### Add attachments and media Insert an Image, Video, Audio, or File block, then choose the file to upload. You can also add page attachments. Use descriptive titles and surrounding explanatory text so another teammate - and the agent - understands why the material matters. ### 3D models Insert a `3D model` block and choose a GLB or glTF file (up to 500 MB). The block shows the model in an orbit-and-zoom viewer with the file name, size and format in the corner. Everything about the block can be changed in place: - `Settings` (the gear in the frame) opens a details table inside the same frame: alt text, a caption shown under the model, auto-rotate, camera controls and height. `Save` writes all of it at once; `Cancel` discards it. The viewer previews the draft while you edit. - `More` offers `Replace model`, `Set AR file` (a USDZ companion so iPhone and iPad can open the model in AR), `Poster from current view` or `Upload poster…` (the still used for PDF export and print), `Remove poster`, `Download` and `Copy link to block`. - `Open in new tab` shows the model on its own page; `Fullscreen` enlarges it in place. A USDZ is an AR companion, not a model: add the GLB or glTF first, then set the AR file. Replacing the model clears the poster and the AR file, because both pictured the old file. Shared pages and exports keep the caption and camera settings. --- ## 4. Build databases Databases make recurring work structured, filterable, and reviewable. Typical examples include content calendars, campaign trackers, research libraries, CRM-style lists, competitor trackers, and task boards. ### Create a database 1. In **Workspace**, choose the add control and select **Database**. 2. Choose whether to create a full-page database or insert one inside a page. 3. Name the database for the operating job, not the tool - for example, `Q3 Content Pipeline` rather than `Content table`. 4. Create the first properties and views before importing a large amount of data. ### Build a useful schema Every database has rows, which open as full pages, plus properties that describe each row. Start small. For a content pipeline, use: | Property | Recommended type | Example | | --- | --- | --- | | Name | Title | `AEO: stop optimizing only for rankings` | | Status | Status or select | Idea, Draft, In review, Scheduled, Published | | Channel | Multi-select | LinkedIn, Blog, Newsletter | | Owner | Person | Content lead | | Publish date | Date | 2026-08-13 | | Strategic theme | Relation or select | Category education | | Source / evidence | URL, relation, or text | Research report link | Available property types can include text, number, select and status, date, person, relation, URL, phone, email, formula, and system fields such as created time, last edited time, created by, last edited by, and ID. ### Choose the right view | View | Best for | | --- | --- | | **Table** | Data entry, audits, research inventories, and detailed comparison. | | **Board** | Workflows grouped by status or another select property. Drag cards between groups to update progress where supported. | | **Calendar** | Publishing schedules, launches, events, and deadline-driven work. | | **Gallery** | Visual assets, creative concepts, and card-based libraries. | Use filters to hide irrelevant rows, sorting to create a decision-ready order, and group-by to make workflow states visible. A database’s shared link respects its saved view configuration, including filtering and order. ### Work in a table - Hover a row to see its controls on the left: **+** adds a row below it (hold Alt to add it above), the grip moves the row when you drag it and opens the row menu when you click it, and the checkbox selects the row. - Choose **Open** in a row's name to read and edit the row as a page. In a full-page database the row opens beside the table, and the table stays editable; **View settings** → **Open pages in** switches between **Side peek**, **Center peek** and **Full page**. - Click a column's header for everything about that column: filter, sort, calculate, freeze the columns up to it, wrap its text, hide it, change its type or edit its options. - The line under the table shows one calculation per column, such as a sum or a count, and keeps it with your view. - Choose **Search** in the view bar to show only the rows that contain every word you type. - Select several rows to change them together: the bar above the table sets a property on all of them, duplicates them or deletes them. Shift-click selects a range; Escape clears the selection. ### Linked and inline databases - Use an **inline database** to keep a small working table inside a planning page. - Use a **full-page database** for a team operating system that needs many views. - Use a **linked database view** to show the same source database in a different page and context. For example, show only LinkedIn drafts on a channel plan page without duplicating the underlying items. --- ## 5. Share and collaborate ### Share a page or database Pages and databases can have share links. Open the **Share** control in the object header, choose an audience, then create and copy the link. To share one **Interactive HTML** block or **3D model**, use **Share this visualization** on that block, choose an audience, then create and copy its link. Visitors see that visualization without the rest of the page. The link shows the block as it is now and stops working when the block is deleted. | Audience | Who can access it | | --- | --- | | **Public** | Anyone who has the link. Use only for material you are comfortable exposing outside your organization. | | **Organization** | Members of the connected organization. This option appears only for organization workspaces. | Changing the audience of an active link rotates its token. The old link stops working, so notify anyone who relied on it. You can also duplicate a public page into your active workspace. Sign in, choose **Duplicate to my workspace**, and edit the independent copy; the source workspace and page stay unchanged. ### Stop sharing and audit live links - Open the object’s **Share** control and choose **Stop sharing** to revoke one link. A visualization has its own link, managed on its block: stopping a page's link leaves its visualization links working, and the reverse. - Open **Settings → Shared links** to review every active link in the workspace, including the object, audience, creator, and creation date. - Use **Revoke all** only when you intend to invalidate every active page, database, and visualization share link in the workspace. ### Send a secure note A secure note sends one private text, such as an access code or a key, to a person who may not have an ActVox account. The note opens once. Then ActVox deletes it. 1. In the sidebar, select **+** and then **Secure note**. 2. Write the text (at most 10,000 characters). 3. Write a password of at least 8 characters, or select **Generate**. 4. Choose when an unread note is deleted: 1 hour, 1 day or 7 days. The default is 7 days. 5. Select **Create link**. Copy the link and the password before you close the dialog. ActVox does not keep them. The recipient opens the link, writes the password and reads the text. No account is necessary. How it works and its limits: - Your browser encrypts the text with a key it derives from the password (PBKDF2-SHA256, 600,000 iterations, then AES-256-GCM). ActVox stores only the encrypted text. It never receives the text, the password or the encryption key. - Send the link and the password on two different channels, for example the link by email and the password in a call. A person who has both can read the note. - The note opens once. ActVox deletes it in the same step that sends it to the browser. If the connection drops during that step, the note is lost. - 5 wrong passwords delete the note. ActVox cannot recover a lost password or a deleted note. - One network can try at most 20 passwords in 15 minutes. One account can create at most 30 notes in one hour. - Connected agents cannot create secure notes. ### Collaborate well - Keep decisions in a page or database record, not only in a chat message. - Use comments and review states for feedback on pages and database rows where available. - Share the smallest useful object. A focused brief is safer and easier to act on than an entire workspace export. - Before creating a public link, check for client data, credentials, personal information, internal planning, and unpublished work. ### Ask ActVox in a comment Where a workspace owner or admin has turned on **Squad → ActVox → Settings → Answer @actvox in comments** (off by default), members can write `@actvox` in a page comment and ActVox answers in the same thread. - Select text and add a comment to ask about that passage. ActVox sees the thread, the quoted passage, the page title, and an excerpt of the page. - The reply reads "ActVox is thinking…" and is replaced by the answer. You get the usual reply notification. If the answer cannot be produced, the thread gets a short notice instead of silence. - Editing the comment you wrote never produces a second answer. - ActVox can do no more than you can. When you ask for a change and can edit the page, ActVox proposes the edit and links the review; nothing changes until a person accepts it. A viewer's question only gets an answer. - Each answer uses the workspace's usage, and a workspace gets at most 20 answers an hour. After that, the thread says the limit is reached and to try again later. ### Squad baseball cards Squad baseball cards show what each team member, person or agent, does well, with the evidence. They are on by default. An owner or admin can turn them off in **Settings → Workspace agent → Squad baseball cards**. - **The card.** In **Squad**, each member shows as a card with their team role, position and name. Open a member and choose **Card** to see the back: their highlights and their monthly batting record. Next to it are all attributes with their scores, the work record and the dots. - **Give a dot.** A dot is one attribute, a score from 1 to 10 and a reason of 10–500 characters. Choose **Give a dot** under a comment or a task result, or on a member's card. The comment or the task is the evidence. Everyone in the workspace sees the dot with your name. The member gets a notice and can reply. - **Your own view.** On your own card, **Add your own view** shows your score as a marker. It never counts in your scores. - **Scores.** Each attribute shows a believability-weighted score: a rater who scores high on an attribute counts more for it. A thin line marks the plain average. An attribute needs 3 dots from 2 people before it shows a score. There is no overall score. - **Compare.** Choose **Compare** on the **Squad** filter row to see every member's scores in one table, one column per attribute. Choose an attribute to sort by it. Choose a person in **Find people like…** to see whose scores are most similar. This shows similarity, not quality. - **The evidence.** Select an attribute on the **Card** tab to see only the dots on that attribute, with people's and agents' averages apart. Choose **All dots** to see every dot again. - **What the team says.** AI writes a short summary of a member's dots at the top of the **Card** tab. Each claim links to its dots. No decision uses the summary. The member, an agent's sponsor, an owner or an admin can ask for it once a day. It is off until an owner or admin turns on **AI summaries** in Settings. It needs 5 dots from 2 members, a new dot since the last summary, and it uses the workspace's usage. The member, an agent's sponsor, an owner or an admin can remove a summary. An org viewer cannot ask. The member or sponsor can answer **Agree**, **Partly** or **Disagree** with a note that everyone sees. - **Big Five.** On your own card, choose **Take the Big Five** and answer 50 statements on one page. ActVox keeps only your five results, never your answers. Members and agents connected to your workspaces can read your results and any outside result. You can take it again 30 days after your last result. If you delete your results, you can take it again at once. You can also add a type from another test, for example MBTI or PrinciplesYou. - **Share a card.** On a card, choose the share icon, then **Create public link**. Anyone with the link sees the picture, name, team role and top strengths, never dots or reasons. Post it to X, LinkedIn or Telegram, or copy the text or the link, or download the image. If a colleague shares your card, you get a notification. The person who shared it, the person on the card, an agent's sponsor, an owner or an admin can turn the link off. - **Attributes.** In the same Settings section, an owner or admin clears the attributes the team does not rate. A cleared attribute leaves every card and **Give a dot**. Its dots stay. - **Rounds.** A round asks the team to rate colleagues for 7 days. In the same Settings section, an owner or admin starts one, or turns on **Monthly rounds** to start one on the 1st of each month. An owner or admin can end a round early with **Close now**. Each member gets 3 to 5 colleagues they share the most work with, and for each one the attributes with the fewest dots. People get a notification and a line at the top of **Squad**. After a round closes, the same Settings section shows who took part. - **Round dots.** Choose **Rate colleagues** to give dots. A round dot needs no comment or task as evidence. To rate anyone else, choose **Rate someone else** in the same sheet. A dot from a member's card does not count in the round. **Who took part** shows each person as took part or not yet, and an agent once it gave a round dot. It is not a score. --- ## 6. Work with the agent The ActVox agent works from workspace context, attached material, relevant pages, and supported integrations. It is designed to create durable results - pages, plans, drafts, records, and artifacts - rather than only returning disposable chat prose. ### Ask good questions Use a clear outcome, relevant context, and a boundary. For example: > Review the competitor notes in this workspace, create a one-page positioning recommendation, and do not change our existing strategy page. You can reference pages, files, and workspace objects in a conversation. Keep source material current so the agent can ground its output. ### Attach files to a conversation Use the attachment control in the composer to add authorized source material to the current request. ActVox uploads the file to durable workspace storage before it sends the message. While an upload is in progress, **Send** and the Enter shortcut remain disabled. Wait until the attachment appears in the composer; if the upload fails, resolve the visible error and attach the file again. The agent can extract bounded text from PDF, DOCX, TXT, Markdown, CSV, JSON, XML, HTML, and other text-based files. Images, audio, and supported video use their media-processing paths. A successfully attached file keeps a persistent serving URL in the conversation, so the message and attachment remain available after reload. Extraction still depends on a readable file: scanned PDFs without text, encrypted documents, corrupt files, and unsupported binary formats may require another format or explicit context in the prompt. Tell the agent what the file is and what to do with it. For example: > Read the attached customer interview notes, identify recurring objections, and create a cited summary page. Do not infer facts that are not present in the document. Do not attach credentials, secrets, or material you are not authorized to process. ### Plan Use **Plan** for work that is multi-step, ambiguous, cross-functional, or consequential. A plan should make the proposed work inspectable before execution. It can include scope, evidence to use, assumptions, steps, expected outputs, risks, and approval points. Plan is appropriate for requests such as: - Create a 90-day content strategy from research and the current brand profile. - Audit competitor positioning and recommend a category narrative. - Design a research and reporting workflow before changing a database. Review the plan, correct assumptions, and approve the next step in the conversation. Do not treat a plan as proof that an action has been completed. ### Act Use **Act** when you want the agent to execute a supported operation. Act can create or update supported workspace objects, organize material, produce reports, generate drafts, and run eligible analysis flows. ActVox verifies required effects before it claims success. For example, when asked to create a page or content draft, the expected result is a real workspace object you can open and edit - not only a pasted response in the conversation. Low-risk, reversible workspace writes may run directly when your request is explicit. Destructive, bulk, publishing, live-document, or otherwise high-impact actions can require review or approval. External availability also depends on integration permissions and product policy. ### Skills Skills are reusable, versioned instruction packages available in a workspace. Find, create and edit them in **Squad → Skills**. Use them from the chat composer’s ✨ menu or through their associated slash commands when available. Use a skill when the task benefits from a repeatable procedure, such as onboarding context capture, creating a company profile, or producing an initial content draft. Good practice: 1. Read the skill description and expected output. 2. Confirm it matches the workspace and intended outcome. 3. Run it with the specific information it needs. 4. Review the generated page, draft, or artifact before sharing or publishing. Skills do not bypass workspace permissions, action allowlists, or approval requirements. User-authored skills are versioned and should not be silently rewritten by autonomous improvement workflows. --- ## 7. Automation and connections ActVox can connect supported sources, platforms, and agent tooling to a workspace. Connections are scoped to the workspace and subject to the connected provider’s permissions, availability, quotas, and policy. ### Connect your AI Open **Agents & API** from the account menu (your avatar at the foot of the sidebar). It has five tabs: **Overview** (the workspace address, the agents and keys, and the last activity), **Connect** (every app and tool you can connect, the social accounts ActVox posts to, and prompts to try), **API keys** (every agent with its connections and keys, and plain keys, with a **Workspace** / **Mine** / **Organization** filter), **Usage** (the requests each agent, key and tool made in the last 7, 30 or 90 days) and **Activity** (each request of the last 30 days, newest first). Choose **Connect an agent**. 1. Pick the AI you use. Some run in your browser (Claude, ChatGPT); others run in your terminal (Claude Code, Codex). 2. For a terminal client, copy the prompt it shows and paste it into the client. It sets up the connection itself and opens an ActVox window. For a browser client, add ActVox to it as the steps show. 3. In the ActVox window, check the agent's name and press **Next**. Choose this workspace and press **Next**. Keep **Full access**, or choose **Read only**, **Read and edit** or **Custom**. Click **Allow full access**, then **Continue**. 4. The flow says the AI is connected. Copy the first task it offers and paste it into the AI. It asks you a few questions, then saves its plan here as a page, and **Open it** takes you there. ### Your agents: names, connections and who can use them Every agent you connect has a card under **Agents & API → API keys → Your agents**, and on its app's own page (Hermes, Codex, Claude Code…). - **An agent** is who it is: a name, an @handle and an app. **A connection** is how it reaches ActVox: *via MCP*, *via the ActVox CLI* or *via an API key*. One agent can have several connections, each with its own **Pause** and **Revoke**. **The ActVox skill** is optional instructions that teach an agent how to work here; it is not a connection. - **Edit identity** on the card opens the agent's profile in **Squad**, where the person who connected it, or an owner or admin, changes its name, @handle and avatar and, for an agent that signs in with the ActVox CLI or a key, which app it is. Click the avatar to upload a picture (PNG, JPEG, WebP or GIF, up to 2 MB), use a link, choose a saved character, or remove it to show the app's logo again; with a picture, the app's logo sits in its corner. Each change shows **Undo** for a few seconds, which puts the previous avatar back, picture included. Comments and tasks follow the new name and face, and the profile's **Name and face** card and the **Activity** tab show every change and who made it. An agent can change its own name, @handle and avatar too, at most 10 times a day. A card that says **Default name · Name it** still carries its connection's name, for example "ActVox CLI". - **Who can @mention and assign it** is one choice: **Nobody** (it takes no mentions or tasks, and turning it off gives its open tasks back to people), **Only me** (only the person who connected it), or **Everyone in the workspace** (only an owner or admin can choose this). - **Missing from a picker?** Check its card: it may be set to Nobody, belong to someone else, be paused or expired, or be approved to read only (reconnect it with **Read and edit**). - When you connect an agent with the ActVox CLI, the approval page asks for its name and which app it is. ### Availability vocabulary The UI uses these exact availability states: `supported`, `canary_passed`, `verified`, `verified_read`, `documented`, `unavailable` and `blocked`. `verified` and `verified_read` are permanent approvals based on real production connections: `verified` means a write was confirmed in production, `verified_read` means reads only. A label says how far ActVox has verified a client end to end. It is not the access level: a connection's permissions are what you approve at consent. The state is documented separately for each product: MCP grants authenticated tools and workspace data; a skill supplies instructions for a local agent. ### AI-agent availability | AI agent | MCP | Skill | | --- | --- | --- | | **ChatGPT** | **verified** | — | | **Claude (hosted)** | **verified** | — | | **Claude Code** | **verified read** | **supported** | | **Codex** | **verified** (through ChatGPT’s connector) | **supported** | | **Hermes** | **verified** through the ActVox CLI | **supported** | | **Grok** | **verified** | — | | **OpenClaw** | **verified** through the ActVox CLI | No catalog skill or installer | Setup steps for each agent are on the [Agents page](/agents#agents-connect). ### Other MCP clients Any other MCP client that signs in with OAuth can connect too, for example an agent platform that asks for an MCP server URL. Add `https://api.actvox.com/mcp` in that app. - The ActVox window marks the app **Unverified** and shows, in large text, the domain that receives the access. Allow access only if you know that domain. - It starts with Full access: read, edit, run actions and share links. Publishing stays off until you tick it. Untick what the app does not need. - The app then has a card under **Agents & API → API keys → Your agents** that reads **Unverified app** and its domain, with **Pause** and **Revoke** like any agent. ### Keep the three stores separate Profile, workspace instructions, and the Knowledge Base are three separate stores. A connection does not merge them into one memory blob. 1. Profile stores the business portrait and identity fields. 2. Instructions store how the agent should work with you. 3. The Knowledge Base stores sources such as sites and documents; indexing uses usage. ### ChatGPT — MCP — verified - Add the ActVox plugin from the ChatGPT plugin directory (https://chatgpt.com/plugins/plugin_asdk_app_6acf1a2ed1d8819182e99b82479a9bff), or press **Add to ChatGPT** on the ChatGPT card in **Agents & API → Connect**. Sign in to ActVox and choose a workspace when ChatGPT asks. - Fallback: in ChatGPT Developer mode, add ActVox as a custom connection with the hosted MCP endpoint `https://api.actvox.com/mcp` and OAuth. - Start with read access for one workspace. ChatGPT has no ActVox skill installer; use the Codex tile for a local skill. ### Claude (hosted) — MCP — verified - In Claude, open Customize → Connectors → **`Add custom connector`**, paste `https://api.actvox.com/mcp`, and choose **`Sign in now`** with Claude’s published OAuth identity. - Team and Enterprise owners can add the connector once in organization settings, but each member still signs in and consents individually. Claude Code is a separate tile. - Approved permanently on 2026-10-03 from production use: live connections wrote and the result was confirmed by read-back, so its setup panel reads **Integration: Tested**. ### Claude Code — MCP `verified_read`; skill supported - The hosted MCP path is approved permanently at **verified read**: live Claude Code connections read workspace data in production, and no write through its own connection has been confirmed yet. Run `claude mcp add --transport http actvox https://api.actvox.com/mcp`, then `claude mcp login actvox` and complete consent. - The installable skill is supported for the `claude` target. Use the Claude Code tile for the installer, key setup, and local verification. ### Codex — skill supported; MCP verified - The installable skill is supported for the `codex` target. Use the Codex tile to install it, issue a workspace key, and run the read-only doctor and capabilities checks. - Hosted MCP is approved permanently: Codex’s production use runs through ChatGPT’s connector and includes writes confirmed by read-back. To connect the Codex CLI with its own identity, run `codex mcp add actvox --url https://api.actvox.com/mcp`, then `codex mcp login actvox` and complete consent. A local skill install is not hosted MCP support. ### Hermes — skill supported; verified through the ActVox CLI - The installable skill is supported for the `hermes` target. Use its tile for the skill installer and workspace-key setup. - Hermes works with ActVox through the skill and the ActVox CLI (a workspace key or a CLI login). That path is approved permanently on production writes, and its setup panel reads **Integration: Tested over the ActVox CLI**; the tile itself reads **Connected** once you connect one here. - Hermes’ own hosted MCP connection stays available: run `hermes mcp add actvox --url https://api.actvox.com/mcp --auth oauth`, then `hermes mcp login actvox` and complete consent. The skill’s instructions do not grant workspace data by themselves; keep the key in the agent’s protected environment. ### Grok — MCP — verified - In grok.com, open Connectors → New Connector → Custom and add `https://api.actvox.com/mcp` with the OAuth values shown on the Grok tile. - Grok has no ActVox CLI or skill installer. The hosted connection is approved permanently: live Grok connections wrote in production and the result was confirmed by read-back. ### OpenClaw — verified through the ActVox CLI; no installer - OpenClaw works with ActVox through the ActVox CLI (the ActVox CLI tile: a CLI login or a workspace key). That path is approved permanently on production writes, and its setup panel reads **Integration: Tested over the ActVox CLI**; the tile itself reads **Connected** once you connect one here. - Install the ActVox plugin from any OpenClaw chat: `/plugins install @actvox/openclaw --accept-capabilities`. Then ask the agent to connect to ActVox. It sends a link, and you approve on your phone or computer. The plugin gives the agent ActVox tools and reads its inbox every 30 seconds. Until you are on the plugin, use the ActVox CLI above: the agent verifies, checks in and says hello with CLI commands, not native tools. There is no catalog skill or installer target for it. - The manifest still lists `local_skill` as a limitation, but OpenClaw has no catalog `skill: true` flag and is not an installable skill client. Do not claim it is supported or add an installer. ### Short first test After a supported skill or an available MCP connection is ready, send this read-only Knowledge Base smoke test: > `Run a read-only Knowledge Base smoke test. Tell me which workspace you can access and list the titles of the first three Knowledge Base sources. Do not write or save any personal information unless I explicitly confirm the exact fields and values first.` This is a smoke test, not the five-step onboarding. If onboarding follows, ask before each write and read back the profile, instructions, and Knowledge Base changes separately. ### Manage and remove - For MCP, **Agents & API → API keys** shows the grant, last call, and verified effect; pause, resume, or revoke access there, then remove the connection in the provider. - For a skill, revoke its workspace key in **Agents & API → API keys** and remove the installed skill from the local agent. - If a consent page says the client is not approved, the current tile policy is authoritative. Retrying does not change a documented, unavailable, or blocked state. ### Your agents and your role - Anyone with the editor role or above can connect their own agent, through a consent page or with a key on a connector tab. You name the agent when you connect it. Only you can @mention and assign it until an owner or admin sets it to **Everyone in the workspace**, and only you see its connections. - An agent never does more than the person who connected it can do now. An editor's agent cannot share, publish or change workspace settings; if you become a viewer, your agent can only read; if you leave the workspace, it stops working. - On Free, Team Starter and Team, AI connectors count for the whole organization: a product another member already connected adds no new connector. ### Automation boundaries An available connection does not mean every action is automatic or approved. Publishing, paid activity, destructive updates, and high-impact external actions can require explicit approval. ActVox does not guarantee external platform availability or campaign outcomes. --- ## 8. Admin and security ### Workspace and organization administration Workspace access follows the authenticated user and workspace or organization boundary. In an organization, administrators can manage organization membership, invitations, and roles in **Squad**: choose **Whole organization** on its filter row. Use the least authority necessary: - Give members access only to the workspace and information they need. - Restrict administrative roles to people who manage membership, settings, or high-impact actions. - Review organization invitations and member roles regularly. - Remove access promptly when a contractor, client engagement, or employment relationship ends. ### API keys Open **Agents & API → API keys** (the **API keys** tab at the top of Agents & API) to create and manage API keys. | Permission | Access | | --- | --- | | `workspace:read` | Read project data. | | `workspace:write` | Modify project data. | | `strategy:read` / `strategy:write` | Read or modify project strategy. | | `content:read` / `content:write` | Read or create and modify content. | | `reports:read` | Read reports. | | `admin` | Full access. Use sparingly. | Prefer a project-scoped key over a multi-project key whenever possible. The secret value is shown only when created. Store it in a proper secret manager or the approved tool environment, not in a page, chat, uploaded file, screenshot, or source code repository. Revoke a key immediately if it may have been exposed, is no longer needed, or belongs to a departing collaborator. You can also delete a revoked key to remove it from the list. ### API keys and agents On the **API keys** page each agent appears once, with how it reaches ActVox (the ActVox CLI, an app or a key), the workspaces it is in and when it was last active. Keys that belong to no agent are under **Plain keys**. **Manage** and **Issue API key** show only for agents you may manage: the person who connected the agent, or an owner or admin of every workspace it is in. - **Manage** opens a panel beside the list with the agent's permissions and workspaces, its keys and connections, and **Revoke agent**. Changing permissions there changes them for every way the agent connects. - **Issue API key** gives an agent a key without touching its other connections: both keep working. The key has the permissions the agent has. - **Replace key** makes a new key with the same name and permissions, then revokes the old one. A key is stored only as a hash, so ActVox cannot show it again: copy the new key, or the `export ACTVOX_API_KEY=` line beside it, while it is on screen. If the old key could not be revoked, the page says it is still active. - **Disconnect** ends one connection. Connecting the same agent again from the same app replaces its older connection instead of adding a second one. - An agent on a server with no saved sign-in can run `actvox auth login --device --agent `. You approve it as that agent, so it keeps its identity instead of coming back as a new one. ### API usage Open **Agents & API → Usage** to see the requests that agents, keys and connections sent to the ActVox API and MCP. - Choose the days, a **key or agent** and a **tool**. The address keeps the filters, so a copied link opens the same view. - The chart shows the requests of each UTC day, split into MCP, CLI and API. A request shows about one minute after it is made. Usage is kept for 90 days. - **By key or agent** lists each agent and your own keys. Select an agent to see its keys and connections. **By tool** lists each tool. Select one to filter by it. - **Refused or failed** counts HTTP 4xx and 5xx answers and rate limits. An MCP tool that answers with an error inside a completed request is not counted. - An owner or admin sees the name of each agent's key and connection. Other members see the agents they can use, and their own keys and connections. The rest show as **Other keys**. The totals are the same for everyone. - Not counted: actions of the chat agent in the app, and requests with a key that is not bound to this workspace. A request through an MCP connection counts in the workspace where the connection was made. - The line at the top shows how much of the plan is used this period. **Open billing** opens Settings › Billing. ### API activity Open **Agents & API → Activity** to see each request that agents, keys and connections sent to the ActVox API and MCP, newest first. - Each row shows the time, the **key or agent**, the **tool**, the **target**, the **status** and how long the request took. Times are in your time zone. - Choose the last 24 hours, 7 days or 30 days, a **key or agent** and a **tool**. The address keeps the filters, so a copied link opens the same view. - The newest page updates by itself. Select **Older** to see earlier requests, and **Newest** to go back. - **Target** is the id of the object that a request named, for example a page or a database row. It shows only next to a key or connection that you can see by name. - The **status** is the HTTP answer: OK, **Refused** (4xx), **Rate limited** (429) or **Failed** (5xx). - An owner or admin sees the name of each agent's key and connection. Other members see the agents they can use, and their own keys and connections. The rest show as **Other keys**, with no target. - Requests are kept for 30 days. A request shows about one minute after it is made. The log keeps ids only: never a key, a token or the content of a request. ### Account and sharing security - Open **Manage account** from the account menu: it opens **Settings → Account**, with your name, email addresses, connected accounts, password, passkeys, active devices and account deletion. Organization admins open **Settings → Organization** from the gear on an organization row: the name, the logo, leaving or deleting the organization, and who can create workspaces. Usage and payments are in **Settings → Billing**. - Invite, change and remove people in **Squad**. **Invite Human** sends an invitation. **Whole organization** on the filter row lists everyone in the organization, the pending invitations, each with **Resend** and **Revoke**, and the agents. Open a person to set their role and their level on each workspace. **Settings → Access** keeps who can open the workspace and the agents' levels. - Use the account security controls for your personal account security settings. - Use **Settings → Shared links** to audit what is externally accessible right now. - Treat public links as public. Do not rely on obscurity of the URL as a security control. - Rotate a link by changing its audience or revoke it and create a new link. - Use organization-only links when the material is intended only for your organization. ### Plan limits Free sets how many blocks your workspaces can hold. Every paid plan has unlimited blocks. Each plan sets how much file storage your workspaces can hold. A personal plan's storage is shared by all your workspaces. On Team Starter, Team and Grow, each paid seat adds to one storage pool for the whole organization: for example, two paid Team seats give it 20 GB. | Plan | Blocks | File storage | | --- | --- | --- | | Free | 1,000 | 1 GB | | Starter | Unlimited | 2 GB | | Personal | Unlimited | 5 GB | | Pro | Unlimited | 20 GB | | Ultra | Unlimited | 50 GB | | Education | Unlimited | 20 GB | | Team Starter | Unlimited | 5 GB per paid seat | | Team | Unlimited | 10 GB per paid seat | | Grow | Unlimited | 20 GB per paid seat | - A block is a paragraph, heading, list item, image, embed, database row or whiteboard element. Empty pages, databases and whiteboards use no blocks. Deleting content frees blocks. - File storage counts every file your workspaces hold: uploads, generated pictures and videos, attachments, kept versions and files in Trash. - AI connectors (Claude, ChatGPT, Grok and other AI products): Free 3, Starter 5, Personal 15, Team Starter 5 per paid seat, Team 15 per paid seat. Pro, Ultra, Education and Grow have no AI connector limit. Other integrations do not count. - The sidebar shows a limit once you use 60% of it. **Billing** shows your file storage against the allowance, and by how much you are over it. --- ## 9. Reports Reports are durable research and audit outputs connected to the workspace. They can capture analysis of websites, markets, competitors, SEO/AEO, social presence, local presence, and other supported business signals. ### Create or run a report 1. Open **Reports** in the Context section of the Workspace OS. 2. Use **Run new analysis** when it is available. 3. State the question, scope, sources, market, or target audience clearly. 4. Review the report’s evidence, assumptions, and recommendations. 5. Turn accepted recommendations into a [Strategy](#10-strategy), a page, a database item, or a task. The standalone reports route redirects to the Workspace OS; the Workspace OS report entry is the canonical place to find and continue report work. ### Review a report critically A report is evidence-informed output, not a guarantee. Before acting on it: - Check the source quality and date. - Separate observed facts from hypotheses and recommendations. - Confirm anything that affects legal claims, pricing, budgets, external commitments, or publishing. - Link the report to the strategy or decision it informed. --- ## 10. Strategy **Strategy** is the durable workspace artifact for the choices that guide execution. It should connect research to priorities, audiences, positioning, channels, campaigns, and measurement. Use Strategy for questions such as: - Which audience segment should we prioritize this quarter? - What category narrative and value proposition should guide content? - Which channels deserve investment, and what will we measure? - What must be true before we launch a campaign? ### Recommended strategy structure 1. **Objective** - the commercial or strategic outcome. 2. **Audience** - who matters most, their context, and the job they are trying to do. 3. **Evidence** - reports, sources, research, and customer signals. 4. **Positioning** - category, differentiated value, proof, and objections. 5. **Strategic choices** - what to prioritize, avoid, sequence, or test. 6. **Execution system** - content, campaign, research, and operating work linked to pages or databases. 7. **Measures** - leading signals, outcome metrics, owner, review cadence, and decision rules. Do not make Strategy a storage place for every idea. Keep it decision-oriented and link to deeper evidence elsewhere. --- ## 11. Design.md **Design.md** is a Context entry for durable design direction. Use it to make creative work more consistent across people, pages, and agent-assisted output. > This guide interprets “Design my” as the Workspace OS **Design.md** context surface. A strong Design.md can include: - Brand personality and visual principles. - Color and typography guidance. - Image, video, illustration, and layout direction. - Examples of what to emulate and what to avoid. - Accessibility and localization requirements. - Channel-specific constraints, such as LinkedIn carousel dimensions or webinar visual standards. Use concise, testable language. For example, write “editorial, precise, high-contrast, documentary imagery; no generic AI blue gradients” instead of “make it modern.” Link approved source assets from [Drive](#13-uploaded) or [Knowledge base](#12-knowledge-base). --- ## 12. Knowledge base The Knowledge base is the workspace source library. Use it for materials your agents should be able to search, cite, and revisit. Uploads elsewhere are not searched until you add them here. ### Add sources You can add: - A web page, with **Add a link**. ActVox reads the page and indexes it. Indexing uses your usage. - A file that is already in **Drive**, with **Add to Knowledge Base**. - A PDF or a Word document (`.doc` or `.docx`) from this page. - Discovered sources from an eligible industry, business type, and location search. Open **Knowledge Base** from **More** (⋯) at the top of the sidebar, or under **Your tools** if you added it there. A ready source reads **Ready** and shows how many sections were indexed. Search the knowledge base to confirm the source is useful. ### Find and use evidence Use Knowledge base search to retrieve relevant workspace sources before writing a brief, report, strategy, or content asset. Source discovery can help identify candidates, but review discovered sources before treating them as authoritative. You can rate discovered sources to improve future discovery relevance. ### Maintain quality - Prefer primary sources, approved brand material, customer research, and dated market evidence. - Remove stale, irrelevant, duplicate, or untrusted sources. - Keep confidential material in the appropriate workspace and never add secrets. - Use meaningful source titles so teammates can assess a result without opening every file. - Re-index an updated source when the source content has materially changed. --- ## 13. Drive **Drive** holds every file in the workspace: what people upload, what the assistant generates, and files attached to pages, posts, chats and records. A file has one **File menu** (⋯, or right-click a card), with the same actions wherever you meet it. Use Drive when you need to: - Keep the original file before turning it into a page or report. - Give the assistant a file to reference. - Find, organize, rename, tag, edit or remove workspace files. - See where a file is used before you change or remove it. ### Add files 1. Open **Drive** in the sidebar and choose **Upload**, or drop or paste files anywhere in the library. 2. To save a file from the web, choose **Add from URL** and paste a public `https://` link to the file. A copy is saved in Drive. 3. Wait for the upload to finish. A failed upload says so; upload it again. 4. Add the file to the [Knowledge base](#12-knowledge-base) when the assistant should search it. A file is not searched on its own. Files count toward your plan's file storage, which all the workspaces on your plan share. The allowance for each plan is in [Admin and security](#8-admin-and-security). ### Find and organize - Filter by type (**Images**, **Video**, **PDF**, **Documents**) or by source (**Uploaded**, **Generated**, **Attached**, **Brand**, **Trash**), or search by name. - Choose **New folder** to group files, then **Move to…** in a file's menu. Deleting a folder keeps its files, at the top level. - Select a file to see its details beside the grid: preview, name, **Alt text**, **Caption**, tags, folder and **Used in**, the places that use it. A replaced file also lists its last 10 **Versions**: download one, or **Restore** it to make it current again. - Select several files to tag, move or trash them together. ### The File menu | Action | What it does | | --- | --- | | **Open**, **File details** | Shows the file, or its details. | | **Edit** | In Drive, and on a picture where it is used. A picture opens the editor over the page, with Crop, Rotate (and flip), Adjust (brightness, contrast, saturation), Markup and Text, and AI edit (describe the change; the editor shows how much of your usage one AI edit takes, and it is returned if the edit fails), with the steps under History: **Save as new version** makes your edit the file's newest version, which every place that uses it shows, and **Save as copy** puts "{name} (edited).png" beside the original instead. A text, Markdown, CSV or code file opens in its editor, and **Save** keeps the previous text as a version. A file too long to open whole, or not UTF-8 text, says why instead. | | **Rename**, **Edit tags** | One name and one set of tags, wherever the file appears. | | **Replace…** | Uploads a new version of the file, of the same kind: a picture for a picture, a video for a video. Every page, database and draft that uses the file shows the new version. A sent message, a published post and a page's history keep the version they showed. | | **Download**, **Download as PNG / JPG / WebP** | Saves the file under its own name, or a picture in another format. Select several files to download them as one .zip (at most 50 files and 200 MB). | | **Copy link**, **Share link…** | **Copy link** copies a link that opens the file for members of the workspace. **Share link…** makes a public, view-only link: choose how long it lasts and whether it allows downloads, and turn it off at any time. Your links are listed in **Settings → Shared links**. | | **Duplicate**, **Move to…**, **Mark as brand file**, **Add to Knowledge Base** | Offered for files that Drive holds itself. | | **Move to Trash** | See **Trash** below. | Where a file is used, its menu adds what that place allows: - **Remove from page**, **Remove from post**, **Remove from message** or **Remove from record** takes the file out of that place only. It stays in Drive. - **Edit**, on a picture, opens the image editor. **Save as copy** puts the edited copy in the picture's place in a post, in a page's image or gallery, and in a record's **File & media** field; the original stays in Drive. Elsewhere (a message, the Content tab's gallery) the copy goes beside the original in Drive. - **Crop for {platform}**, on a post's picture, saves a cropped copy in the platform's shape and uses it in the post. - **Show in Drive** opens the library on the file. - In **Content**, the **Gallery** view shows generated pictures. A generated picture leaves the gallery seven days after it was made unless a post uses it; **Keep in Drive** saves it to the library first. ### Trash 1. **Move to Trash** asks first when the file is used somewhere, and **Undo** brings it back right after. 2. Where a file in Trash was used, a placeholder says so, with **Restore file**. 3. In **Trash**, choose **Restore** to bring a file back, or **Delete forever…** to delete it. It is deleted 30 days later, and until then you can still restore it. The assistant and connected agents can move files to Trash and restore them. Only a person can delete a file forever. Files are not listed publicly. Only a **Share link…** opens a file for people outside the workspace, until you turn it off or it expires. --- ## 14. Memory **Memory** holds short facts about this workspace. You add them with **Remember**. The chat and a connected agent can add them too. Facts the chat stored appear under **Remembered in chat**. A fact that is only yours is marked so. **Forget** removes one you are allowed to remove. Examples of useful memory: - A priority audience, a positioning constraint, or a target region. - An approved terminology rule or a known objection. - A recurring operating preference that should shape future work. Memory is workspace-scoped. Treat it as short context, not as a secret vault and not as a substitute for a page or a source. ### Use memory responsibly - Put standing operating instructions in **ACTVOX.md**. - Put files and web pages agents should search in the [Knowledge base](#12-knowledge-base). Keep the original file in [Drive](#13-uploaded). - Keep a decision in **Strategy**, a report, or a page where it can be reviewed and revised. - Do not rely on memory alone for a critical claim. Confirm it against a current source. --- ## 15. Conversations Conversations are the agent work surface. They live in the **Agents** area of the Workspace OS and keep the discussion connected to the workspace’s context and objects. ### Use conversations for work, not just questions A productive conversation usually has: 1. **Outcome** - what you need delivered. 2. **Context** - relevant pages, sources, files, or business constraints. 3. **Boundary** - what the agent may and may not change, publish, or assume. 4. **Review point** - when you want a plan, draft, recommendation, or approval request. Example: > Plan a competitor positioning review for our UK market. Use the knowledge-base sources and the current strategy. Create a page with evidence and recommendations, but do not alter Strategy until I approve it. ### Manage conversation output When a conversation produces something worth keeping, ask the agent to save it as a page, report, strategy update, content draft, database item, or artifact. Open the resulting object and review it outside the chat. This creates a durable record and makes the work reusable by the team. Use attached files, referenced pages, and skills deliberately. They improve grounding but do not remove the need for human review. --- ## 16. Studio **Studio** is where marketing work is made. It is four tools: **Content**, **Calendar**, and **Brand DNA** in the sidebar, and the **Media studio** inside a draft. A draft is a post for one channel, with its images or video. ### Content **Content** is the output area for marketing assets such as drafts, adaptations, and media-oriented work. It is designed to connect content back to workspace strategy, research, and source context. Use Content to: - Create and review draft marketing assets. - Organize variants and adaptations. - Link content to the relevant audience, campaign, channel, strategic theme, and evidence. - Prepare work for a content calendar or publishing workflow. ### Content workflow 1. Start with a strategic intent and a specific audience. 2. Give the agent approved sources, positioning, and channel constraints. 3. Ask for an editable draft - not automatic publication. 4. Review factual claims, voice, differentiation, CTA, and compliance. 5. Save the approved draft in Content and connect it to a campaign or database record. 6. Use the relevant integration or publishing process only after the required approval. ### Draft, adapt, and schedule a post The fastest path from an idea to a scheduled post runs through the agent and a calendar database. 1. Ask the agent for a draft and name the platform, the audience, and the goal. 2. Ask for adaptations of the approved angle for the other platforms you use. 3. Review each draft in Content and edit the claims, voice, and call to action. 4. Set the status to Scheduled and pick the publish date on the linked calendar record. 5. Publish through the connected account only after the required approval. > Example request: "Draft a launch post for Instagram for local cafe owners, then adapt it for LinkedIn and X, and put both on next week's calendar." ActVox can generate and adapt content, but it does not guarantee reach, rankings, engagement, conversions, or revenue. For video, open a content draft and choose **Generate Video**. In the **Video** lane of **Media studio**, the **Brand style** switch leaves the saved brand's style out of that one request without changing the saved brand. The switch appears when a measured brand is loaded. ### X threads, X Articles, and Substack posts A draft can hold an X post, an X thread, an X Article, or a Substack post in the shape the platform itself uses. The preview shows the piece as it will look after publication. 1. Open **Content** and start a new draft. Choose **X thread**, **X Article**, or **Substack**. 2. Write the draft in its editor. A thread has one card per post. An X Article and a Substack post have a rich editor with only the blocks that platform supports. 3. Check the preview next to the editor. A block that the platform does not support shows as an unsupported block. 4. To publish an X Article or a Substack post yourself, use **Copy for X** or **Copy for Substack**, paste into the platform's editor, and publish there. Then check the items under **Check after pasting**, such as images, code, and the paywall. **Generate with AI** in the new-draft dialog also writes an X thread, an X Article, and a Substack post. Select the formats, enter a topic, and select **Generate with AI**. A thread comes back as posts of up to 280 characters each. An X Article and a Substack post come back as markdown with a title heading. You can also ask the agent in chat, for example: "Write an X thread about our pricing change." ActVox makes the draft only. It does not post to X or Substack. A publisher agent can read the steps for a draft from `GET /api/v1/workspaces/{workspaceId}/content/{contentId}/publish-request`; nothing is sent by that call. The Substack shape is unofficial, because Substack has no public API. ### Calendar Open **Calendar** to see when each draft goes out. **New content** there starts a draft. Approved posts publish to Instagram or LinkedIn once that account is connected. ### Brand DNA **Brand DNA** is the colours, type, logo, and voice that every image and draft follows. Open it in the sidebar. **Scan my brand** reads a website or a supported social profile and fills those fields. Reference sites are a separate list: they steer composition and do not change the palette or the logo. ### Media studio Open a draft to edit its images, a carousel, or video. Image, Carousel, and Video are modes of that one editor. Generated images and video also appear in **Drive**. --- ## 17. Artifacts **Artifacts** are generated workspace outputs that are not simply a page or content draft. They can include results from supported agent or media-generation workflows and retain the context needed to inspect or reuse the work. Use Artifacts to: - Review a generated deliverable before deciding where it belongs. - Keep output provenance connected to the workspace. - Promote a usable output into Content, a page, a report, or another operating record. - Separate experimental or generated material from final approved source-of-truth documents. ### Review before reuse Before sharing, publishing, or relying on an artifact: - Open it and inspect the actual result. - Check that it matches the requested audience, brand, source, and channel constraints. - Verify factual assertions against the original evidence. - Confirm rights, permissions, and platform requirements for any external use. - Move or link the approved result into the page, database, Content item, or campaign where the team will operate from it. --- ## Operating principles 1. **Context first.** Keep the business profile, instructions, evidence, and strategy current. 2. **Pages for thinking, databases for repeatable operations.** Do not turn every page into a table or every table into a document. 3. **Plan before consequential work.** Review scope and assumptions before execution. 4. **Act with review.** Inspect created objects and approve high-impact external activity explicitly. 5. **Preserve evidence.** Link recommendations to reports, sources, and decisions. 6. **Share narrowly.** Prefer the smallest audience and the smallest object that solves the collaboration need. 7. **Protect secrets.** Do not place credentials, tokens, or sensitive personal data in public links, pages, prompts, or uploaded files.