Skip to main content

​
Getting Started

No. Sign in at agent.creao.ai and start chatting immediately. Workspaces are an optional way to organize your agents and files as your usage grows.
Yes. Sign up for free and start using the super agent right away. The Free plan includes the core super-agent experience; plans differ in credits, support, access to premium models, and a small number of resource-intensive generation capabilities.
New CREAO accounts start with their own team, named from the first name or email prefix you sign up with — you can rename it at any time on the team page. Your plan powers that team: invite people for free and they share its credits and paid features, with per-member limits you control. Using CREAO solo doesn’t change anything; the team is simply there when you want to bring someone in. (If you joined by accepting a team invite, your team is the inviter’s instead.)
Anything you’d ask a technical teammate: generate a web app, analyze a dataset, draft a report, scrape a website, create charts, send emails, build PDFs, automate workflows, and more. The super agent runs code in a full Linux sandbox, so it can install packages, call APIs, and produce real files — not just text responses.

​
Agents

An agent is a reusable workflow built from a successful chat session. Instead of re-prompting the same task, you save it as an agent with structured input fields. Anyone on your team can run it with different inputs, and every run is tracked with version history.
After a successful chat session, click Create Agent in the chat toolbar. The super agent analyzes the conversation and extracts the workflow into a reusable agent with a structured input form, a skill definition, and version control.
Yes, on any paid plan (Pro or higher). Open any agent, go to the Schedule tab, and configure a recurring schedule — daily, weekly, monthly, or a custom cron expression. Scheduled runs execute autonomously and consume credits like manual runs. On Free, existing schedules stay saved but don’t run until you upgrade.

​
Skills and Connectors

Skills are instruction packages that teach the super agent how to perform a task (like a recipe). Connectors give the super agent access to external services via MCP (Model Context Protocol) — Gmail, Slack, Google Sheets, GitHub, and more. Skills define behavior; connectors provide integrations.
Yes, three ways: install from a public GitHub repo URL, upload a ZIP file containing a SKILL.md, or ask the super agent to create one directly in chat. Custom skills use a simple markdown format with YAML frontmatter.
CREAO supports a growing connector catalog including Gmail, Google Calendar, Google Sheets, Google Docs, Google Drive, Google Tasks, Google Ads, Google Analytics, Google Search Console, Outlook, Microsoft Teams, Asana, Linear, Notion, Slack, Discord, Facebook Ads, Reddit, YouTube, GitHub, Semrush, Perplexity, Firecrawl, X, Telegram, Shopify, eBay, and gated previews such as Etsy. Check the Skills and Connectors page for the latest list.
A connector can access only the data allowed by the permissions (scopes) you approve with that provider. Access is scoped to the connected account and the connector’s authorized capabilities. See Connector Security for an overview.
Yes. You can disconnect a connector in CREAO at any time. For complete revocation, you can also remove CREAO app access from the provider’s own security settings.
When you connect your own AWS account (Bring Your Own Cloud), CREAO does not store long-lived cloud credentials. You deploy a CREAO-provided CloudFormation stack in your account that creates a cross-account IAM role; CREAO assumes it on demand using short-lived STS credentials gated by a per-connection ExternalId. The agent reads only what the tool you triggered returns — read-only CloudWatch logs, metrics, and alarm states for debugging. Those tool results flow into the agent’s context, which means they are sent to your selected LLM provider as part of the conversation (the same as other connector data); CREAO does not bulk-export or continuously ingest your cloud account. You can disconnect at any time (which also deletes the stored connection), or delete the CloudFormation stack in your own account. See Privacy and Cross-Account Cloud Access (BYOC) for details.
Your device runs a small daemon that registers its available capabilities (like running shell commands, transferring files to or from the device, or a custom MCP server you configure) and receives lightweight task notifications over AWS IoT Core, then fetches the durable task over an authenticated connection. Each task is checked against your device’s local consent policy — allow, deny, or ask (an interactive prompt in the desktop app, which can also be answered from CREAO on the web or your phone). New pairings default to ask for the shell-command capability, so nothing runs unattended until you decide otherwise. Allowing a Local Worker to start can also allow its tools for that run. Long-running tasks continue without keeping the chat open, show durable status and deadlines, and can be stopped from their task card. A vision-capable Local Worker sends screenshots of the apps it operates to the Lite vision model’s providers, including again when the agent continues a finished worker; its sessions stay on your Mac and are not uploaded to CREAO. See AI & Model Usage. You can pause or revoke a device at any time. See Device Connect Security for details.
Built-in skills are instruction packages, not standalone third-party services. Data is shared externally only when required by model providers or connected tools used during execution. See Skill Security.
They run under the same platform security boundaries, but their trust model differs. Built-in skills are curated by CREAO, while custom skills are user-provided instructions and should be reviewed before enabling.

​
Data Privacy

Person-to-person private chats are readable only by the two participants. Room guests can participate in their room’s private chats. Team owners and admins have no oversight access, and these chats are not sent to AI models or included in room search or summaries. If you chat privately with a room agent, your messages and room context go to that agent’s model. The teammate who added the agent can review your messages and its answers in the agent’s runs. Those run copies follow that teammate’s run-history retention and are not removed when you delete a private message or your account. When someone leaves the room or team, the other participant keeps read-only history. Unsent uploads expire after 24 hours; failed file deletions are retried, and deleted file versions are scheduled for removal after one additional day. See Private chat data and retention.
Where this is enabled, your personal agent can attend a Zoom, Google Meet or Microsoft Teams call to keep everyone in it on the same page: you paste a meeting link (“Send to a meeting”) or ask it to join one in your conversation with it, or turn on “Join my meetings” in its settings, in which case it joins only the calendar events you organize, never invitations from others, and you can tell it to skip a call. Each call gets one agent: if another person’s agent is already in it, yours is not sent. Before the call it gathers context from the calendar event and from your connected documents, messages and email (for example Notion, Google Docs, Google Drive, Slack, Microsoft Teams, Gmail and Outlook) and copies relevant excerpts, with a short brief, onto a context page for the meeting. Before a calendar meeting it tells you what it gathered, and you can remove items or keep the material out of the meeting. A bot operated by Recall.ai joins as a participant under the agent’s name and tells the call in the meeting chat that it is an AI agent transcribing the meeting for the organizer: for a meeting from your calendar in its one opening message with the brief and the context page’s link, posted as the call begins; for a call you send it to by link in a first message saying it is gathering context, with the context page’s link, followed by the brief, what else is on the page and its link once the material is gathered, usually within a few minutes. That chat message is the only notice the other participants get, and the first seconds of a call can be transcribed before it appears; from then on anyone with the link can read the page without signing in, including people who are not CREAO users, so material from your sources that you did not remove or keep out, which can include other people’s messages and emails, is visible to them. During the call the page also shows the names of the people seen in it and the latest lines of the transcript, and after the call it holds the meeting’s record: a summary, what was decided, what is still open and the next steps. Anyone with the link can read and download the whole transcript, and you can copy the link from the meeting page to share it with people who were not in the call. The page, its transcript and its link expire 3 days after the call ends. During the call the agent answers in the chat at once when it has something to say (a short quick reply written by DeepSeek V4.1 Flash, as described in Privacy), reacts to the line it heard with an emoji on its video tile, and, when something needs looking into, looks it up in the background in the meeting’s material and on the web (and, if you chose “everything” in its settings, in its memory of you, your files and the connected services you leave on for that call on the meeting’s page, where anything it would change or send waits for your approval), follows up in the chat with what it found, keeps on the page what is worth keeping for the meeting, and marks where people disagree (who holds which position, and the source behind each) and what is decided (with its own note on risks or conflicts with the meeting’s sources). The bot never speaks. Its video tile shows a page of what it is doing, what it has added and what is still to settle, drawn from the context page, and it writes only in the meeting chat, which every participant can read. CREAO does not record audio or video. Recall.ai processes the call’s audio, chat messages and participants’ display names in the United States, and keeps no recording and no transcript after the call. It streams the call’s audio to ElevenLabs, which transcribes it in whatever language is spoken, under the ElevenLabs terms described in Privacy. CREAO holds the transcript lines (speakers’ display names and what was said or typed, including by people who are not CREAO users) for the length of the call, at most 8 hours. When the call ends, a copy is kept with the page until it expires, and the transcript, headed by the page’s link and the names of the people seen in the call, goes to your personal agent, which writes a summary in your own conversation with it; that copy is removed 30 days after it is delivered, and the summary follows normal conversation retention. Only you see the summary. With the “everything” setting, if your Gmail or Outlook is among the services left on for the call and the attendees’ email addresses are known, the agent drafts one recap email to them, which waits for your approval on the meeting page and is never sent without it. To remove the bot, anyone who can remove participants from the call can remove it, and you can take it out yourself. It leaves by itself about a minute after you leave the call and do not come back, about 30 seconds after everyone else has left, when the call ends, after 10 minutes waiting to be admitted, or after 4 hours at the latest. See Privacy for what is sent to AI models before and during a call and for calendar data.
Being removed revokes your access to the team workspace, but does not automatically delete your files there. Team deletion also does not automatically erase each member’s private workspace files. There is currently no automatic deletion deadline for these files after departure. To request erasure, contact privacy@creao.ai. See Data Retention for details.
CREAO does not train its own models on your conversations, files, or data. Anthropic, OpenAI, Google, MiniMax, Z.AI, Moonshot AI, Thinking Machines, DeepSeek (open weights), Fireworks AI, Meta (Muse Wonder, Muse Ember and Muse Spark models via the Meta Model API), and BytePlus are accessed under API agreements or provider API terms that prohibit training use. MiniMax M3 conversation requests are served by Fireworks AI under CREAO’s Fireworks account (Fireworks’ API terms prohibit training use) — if Fireworks returns a capacity error, the request is automatically retried via OpenRouter and served by DeepInfra, Novita, or Together AI under OpenRouter’s API terms, which prohibit training use — while MiniMax’s own API serves MiniMax M3.1 Flash conversation requests directly (including attached images) and CREAO’s internal response-quality evaluation, under MiniMax’s API terms, which prohibit training use; Moonshot AI, Thinking Machines’ Inkling and Qwen3.8 Omni Flash (served by Alibaba Cloud, the only host OpenRouter offers for it, so its conversation content and attached images reach OpenRouter and Alibaba Cloud) are reached via OpenRouter, whose API terms prohibit training use (Kimi K3 is served by Fireworks AI under CREAO’s account, and images attached to a Kimi K3 conversation reach OpenRouter and Fireworks; Kimi K3.1 Pro is served by Moonshot AI itself, the only host OpenRouter offers for it, so its conversation content and attached images reach OpenRouter and Moonshot AI), and GLM 5.2 / Moonshot AI requests may be served by Fireworks AI under CREAO’s Fireworks account — Fireworks’ API terms also prohibit training use. GLM 5.3 and GLM 5.3 Flash are the exceptions among the Z.AI models: they are sent directly to Z.AI under CREAO’s own Z.AI API agreement, which prohibits training use, so by default their conversation content reaches Z.AI and no other company — not even a relay. CREAO can also route GLM 5.3 (not GLM 5.3 Flash) to Fluxion, which serves it from its Asia region under CREAO’s Fluxion account; this is not the default route, and Fluxion is not covered by this no-training statement until CREAO confirms its data-use terms. (GLM 5.2 is still reached via OpenRouter. CREAO can move the GLM 5.3 models onto that path as a fallback, in which case OpenRouter relays the request and Z.AI still performs the inference.) Muse Spark 1.3 and 1.2 go directly to Meta by default, and reach OpenRouter only if CREAO moves them onto that path as a fallback, in which case OpenRouter relays the request and Meta still performs the inference, being the only host OpenRouter offers for those models (Muse Spark 1.1 has no such fallback). Muse Wonder 1.0 and Muse Ember 1.0 take the same path as Muse Spark 1.3, with the same OpenRouter fallback. DeepSeek V4.1 Flash is served by default through OpenRouter, preferring Fireworks AI under CREAO’s Fireworks account. Conversation messages, context, and attached images reach OpenRouter and the serving inference host; OpenRouter may use another host if Fireworks is unavailable. DeepSeek’s own API is excluded from Flash routing. These paths prohibit training use under their API terms. CREAO can also route Flash to Fluxion, which serves it from its Asia region under CREAO’s Fluxion account; this is not the default route, and when CREAO uses it, conversation messages, context, and attached images reach Fluxion instead of OpenRouter. The no-training statement above does not cover Fluxion until CREAO confirms Fluxion’s data-use terms. Selecting Flash does not keep content exclusively on CREAO-managed systems; the self-hosted V4 checkpoint is a separate, explicitly selected legacy route. DeepSeek V4 Pro uses a different serving provider: it is NOT self-hosted — every conversation request for it is relayed via OpenRouter and served by DeepSeek’s own API by preference (GMICloud as a fallback), so conversation content for that model IS sent to DeepSeek, under DeepSeek’s and OpenRouter’s API terms, which prohibit training use. Fugu Ultra via Sakana AI is a qualified exception while governed by Sakana’s public Fugu terms: submitted content may be used for Sakana training or improvement unless an opt-out or separate no-training/DPA arrangement is in place, so do not use Fugu Ultra for regulated, sensitive, confidential, or personal data without that confirmation. Team-room routing sends a room post that @mentions no agent, with recent room context, to TypeSafe’s Jev decision model via OpenRouter, which returns scores only; TypeSafe’s published privacy policy says it does not train or fine-tune models on that input, keeps it as long as reasonably necessary to provide its service rather than for a fixed period, and processes it in the United States.Music prompts (including lyrics) and generation settings are sent to ElevenLabs when music generation is enabled; selected Scribe transcription requests also send audio to ElevenLabs. Its training and retention terms differ from the API commitments above. See the ElevenLabs disclosure in Privacy; CREAO does not promise zero retention for music.
Data is stored in cloud infrastructure and protected with encryption in transit and at rest. Execution runs in isolated sandbox environments. Team-room routing requests sent to TypeSafe’s Jev decision model are processed by TypeSafe in the United States.Team-room human requests also store the agent’s ask and your answer. Requests expire after 14 days and are deleted 90 days after closure; account erasure removes requests you own or settled and your recipient records. See Data Retention.ElevenLabs uses US storage by default, but its international affiliates and subprocessors may process content elsewhere. Provider retention and transfer terms are described in Privacy. Where enabled, hourly room summaries send recent room posts, member names, attachment file names, choices, the previous summary, and room task, plan and schedule state (including titles, assignees, due dates, progress and counts) to the DeepSeek V4.1 Flash path described above, even when nobody opens the overview. CREAO pays for this processing. Summaries and up to three suggested next steps, including their request text, are stored as room messages, visible to room members and room and team owners and admins, with the same retention and deletion. Personal overview pins are private to each member and remain until unpinned, the account is deleted, or the room or membership record is permanently deleted; leaving or archiving a room does not delete pins. Only summary-message metadata and pin identifiers are exported to CREAO’s analytics warehouse; summary text and suggested next steps are excluded. See Privacy for details.
CREAO estimates your approximate location from your IP address at login. A country code is used for analytics, abuse detection, and billing routing, and CREAO may also derive a coarse city/region estimate, used for regional analytics and to segment location-relevant CREAO marketing and lifecycle emails by area. This estimate is approximate — it reflects your network, not a precise position, and can be wrong (VPNs, mobile carriers). CREAO does not collect GPS or precise-location data, and only the coarse city/region estimate — never your raw IP — is shared with our marketing-automation provider (customer.io) for this. Because CREAO does not yet operate a per-user consent mechanism for location-based marketing, users in the EU/EEA/UK are excluded from it entirely. You can opt out of marketing emails at any time via the unsubscribe link or your notification settings. See Privacy for details.
Yes. You can delete conversations, files, and memory entries, and request CREAO account deletion. Where memory stores are enabled, ask the agent to forget a memory it saved; deleting an agent deletes the memories it kept. A limited audit record of a deletion request itself may be retained where required for legal, compliance, and security purposes, as described in the Privacy policy. A Developer Platform account created at developer.creao.ai is a separate product account and is not removed by CREAO account deletion; request that erasure at privacy@creao.ai.One deletion request can also be initiated automatically: if you joined CREAO only by accepting a team invite (an account created without a workspace of its own) and you later leave your last team, CREAO files an account-deletion request on your behalf, since the account has no remaining purpose. Like any deletion request it is reviewable and reversible before it is finalized.If agent alignment is enabled in the preview, the companion can review its own conversations with you, your feedback, its actions and the room replies to its posts that you can access with its model provider, then propose learned rules from your alignment-card answers. A separate confirmation shows the exact replacement text before saving. Rules apply across the companion’s later chats and rooms and may contain names or paraphrases; room headings are not access restrictions. Your room replies may also reach another member’s companion model for their alignment review. Learned rules have no automatic expiry. Agent or team deletion does not erase these files. Account hard deletion removes current files. Historical versions may remain after replacement or deletion until a separate verified purge; do not rely on automatic expiry during this preview. Contact privacy@creao.ai for historical erasure. Deleting your account does not remove names or paraphrases already saved in another person’s learned rules. Contact privacy@creao.ai to request erasure of those copies. Alignment cards save quoted evidence, including other members’ room replies, in the owner’s home conversation; those copies follow normal conversation retention and existing analytics warehouse handling. Deleting the source member does not erase copies in another owner’s conversation; request their erasure at privacy@creao.ai.Every turn in your personal companion’s home conversation is also stored as a run record with the first 500 characters of your message, whether or not alignment is enabled. These records follow existing run/session retention and analytics warehouse replication.If the personal companion preview is enabled, deleting its team disables the companion and erases its encrypted soul notes and remembered facts, but retains the rest of its configuration (including onboarding profile, home-conversation identifier, daily check-in settings, and whether you used CREAO before it and closed its walkthrough) in the database and analytics warehouse, with no automatic hard-purge deadline. Team deletion does not erase this data; home and delegated conversations follow normal conversation retention. A remembered fact you correct keeps its earlier text for 90 days; facts kept before they moved to its memory store are no longer shown in its menu and stay stored until its team or your account is deleted. Submitting an account-deletion request does not itself erase companion data. The administrator-performed account hard delete removes user-owned companion records, soul notes, remembered facts, daily check-in settings, suggested changes to your agents (which team deletion keeps with the agent’s record) and conversations, and removes your agents from other teams’ rooms where it can (a room it could not clean is recorded), whose messages stay as the rooms’ history; warehouse replicas follow source deletion. Contact privacy@creao.ai for an erasure request or about anything that remains.
CREAO is designed around core privacy rights including access/export and deletion. Enterprise customers can request a DPA. See Privacy for details.
Yes. The CREAO iOS and iPadOS app (Apple App Store) shows two notices that web and Android do not:
  1. Before sign-in — you must accept CREAO’s Privacy and Personal Information Protection Agreement before creating an account or signing in.
  2. After sign-in — before using AI features, you must acknowledge that your messages, files, and other content may be sent to third-party AI providers (such as Anthropic and OpenAI) for processing.
You can review the full policy anytime from Settings → Privacy. See Privacy for details.
Secrets are encrypted at rest and injected at runtime for execution. Values are not shown in plain text after creation, except through whitelisted and logged campaign audit exports used by authorized admins to verify trading activity. Trading-campaign secrets, including WEEX API credentials and the S2 debot developer API credentials (DEBOT_API_KEY / DEBOT_API_SECRET), are not stored in campaign tables. S2 never collects a wallet private key — debot’s custodial wallet API signs Robinhood Chain transactions server-side. In a team room, room owners and admins can add room secrets: room members from the team see their names (guests don’t), never their values, and every agent running in that room receives them, so anyone who can message those agents can make them use one.

​
Sandbox and Environment

Your sandbox stays active for the duration of the conversation. Installed packages, created files, and running processes persist between messages. After 30 minutes of inactivity, the sandbox pauses automatically — everything is saved and resumes in about one second.
Yes. Every chat thread gets its own sandbox. Work done in one thread never affects another. Agent runs also get isolated sandboxes, created from the agent’s saved environment snapshot.
Yes. Add your API keys as Secrets from the Secrets page. They’re encrypted at rest and injected into the sandbox as environment variables (process.env.YOUR_KEY). The super agent can use them in code execution.

​
Memory and Context

Where agent file memory is enabled, an agent learns job-specific rules, procedures and reference facts from its creator’s conversations and automatic runs. Authorized members can view this memory in the agent’s Memory tab, and it can be used in their runs. Other members’ chats and room conversations do not teach it. Personal facts are not intended for shared agent memory; avoid teaching an agent information that should remain private. Delete the agent to queue deletion of its memory files, or contact support about account deletion. Storage cleanup runs asynchronously with retries.
Yes. Memory is cross-thread — when the super agent learns a preference, fact, or decision, it saves it and recalls it in future conversations. You can view, search, and delete memories from Settings → Personalization. Memory is curated when you leave a conversation; for accounts in the segment-curation experiment, a long chat is also curated in segments, after an hour of inactivity or every 24 of your messages. Where memory stores are enabled, the super agent also keeps short memories of your work (decisions, outcomes, commitments and facts about you) in a private store for each workspace you use, saving them without asking each time, and an agent you create keeps the decisions you settle with it in a store of its own, private unless the agent judges one to be for the team you share it with, such as a team policy; ask the agent what it remembers, or to forget something. Your personal agent can read what your own agents keep, and its settings list the memories it keeps for you, to delete. Ask it to forget or correct a memory at any time; see the Privacy policy for how long they are kept.
Memory persists across all conversations indefinitely — use it for durable facts like preferences and decisions. Thread notes are scoped to a single conversation and help the super agent maintain context during long sessions.
Dream is an automatic profiling feature that consolidates your conversation history into a concise portrait of who you are — your role, preferences, working style, and recent decisions. Like the brain during sleep, Dream replays your past conversations offline and distills patterns into a structured profile. This profile is injected into every new conversation so the super agent already knows you from the first message. You can view your Dream profile, trigger a manual regeneration, and browse previous versions from the Memory → Dream tab.

​
Pricing and Limits

CREAO offers five plans. Every plan includes the core super agent; plans differ in credits, support, premium or gated models, and a small number of resource-intensive generation capabilities.The Free plan requires no credit card. Every plan includes the core AI super agent — memory, code execution, web search, custom agents, and built-in, custom, and open-source skills. Free includes 1 personal connector per agent app. You can connect more than one account, but combining personal connectors in one agent requires a paid plan. Paid plans add more credits, higher support levels, and premium models. Pro Plus and above also add scheduled runs and durable long-running video and audio renders, with each higher tier raising how many scheduled runs and renders can run at the same time.
The core product works on every plan, including the free tier — chat, memory, skills, one personal connector per agent app, custom agents, and the API Trigger (programmatic agent runs via API keys) all run on credits from your balance. A few distinctions are paid-only: premium models — the top-tier Claude models (Fable and Opus), Fugu Ultra, and the premium video models (Veo 3.1, Gemini Omni 1.1 Flash, and every Seedance 2.x model, which include all the models that can edit an existing video) — require a paid plan, and free-plan users pick from simplified model tiers rather than hand-picking a specific model. Some preview or entitlement-gated models are available only to eligible accounts. Scheduled runs and durable, long-running video and audio renders require Pro Plus or higher — Free and Pro schedules stay saved but don’t execute until you upgrade, and each tier from Pro Plus up raises how many can run at the same time. Generated PDFs longer than 10 pages also require a paid plan; quick renders and PDFs up to 10 pages remain free on every plan. Video generation on the Free plan is limited to one video per rolling 30-day period; on paid plans, additional video generations simply consume credits from your balance. See Models.
Yes. You can purchase credit packs at $0.15 per credit from the pricing page, on top of your plan’s monthly allocation.
CREAO supports multiple models including Claude (Fable, Opus, Sonnet, Haiku), Gemini, GPT variants, GLM 5.2, Kimi K3, MiniMax, Muse (Wonder, Ember and Spark), Qwen, and Fugu Ultra. Generation-focused models include Veo (video) and BytePlus Seedance / Seedream (video and image). You can select your preferred model from the model selector in the chat interface.

​
Data Privacy

Person-to-person private chats are readable only by the two participants. Room guests can participate in their room’s private chats. Team owners and admins have no oversight access, and these chats are not sent to AI models or included in room search or summaries. If you chat privately with a room agent, your messages and room context go to that agent’s model. The teammate who added the agent can review your messages and its answers in the agent’s runs. Those run copies follow that teammate’s run-history retention and are not removed when you delete a private message or your account. When someone leaves the room or team, the other participant keeps read-only history. Unsent uploads expire after 24 hours; failed file deletions are retried, and deleted file versions are scheduled for removal after one additional day. See Private chat data and retention.
Where this is enabled, your personal agent can attend a Zoom, Google Meet or Microsoft Teams call to keep everyone in it on the same page: you paste a meeting link (“Send to a meeting”) or ask it to join one in your conversation with it, or turn on “Join my meetings” in its settings, in which case it joins only the calendar events you organize, never invitations from others, and you can tell it to skip a call. Each call gets one agent: if another person’s agent is already in it, yours is not sent. Before the call it gathers context from the calendar event and from your connected documents, messages and email (for example Notion, Google Docs, Google Drive, Slack, Microsoft Teams, Gmail and Outlook) and copies relevant excerpts, with a short brief, onto a context page for the meeting. Before a calendar meeting it tells you what it gathered, and you can remove items or keep the material out of the meeting. A bot operated by Recall.ai joins as a participant under the agent’s name and tells the call in the meeting chat that it is an AI agent transcribing the meeting for the organizer: for a meeting from your calendar in its one opening message with the brief and the context page’s link, posted as the call begins; for a call you send it to by link in a first message saying it is gathering context, with the context page’s link, followed by the brief, what else is on the page and its link once the material is gathered, usually within a few minutes. That chat message is the only notice the other participants get, and the first seconds of a call can be transcribed before it appears; from then on anyone with the link can read the page without signing in, including people who are not CREAO users, so material from your sources that you did not remove or keep out, which can include other people’s messages and emails, is visible to them. During the call the page also shows the names of the people seen in it and the latest lines of the transcript, and after the call it holds the meeting’s record: a summary, what was decided, what is still open and the next steps. Anyone with the link can read and download the whole transcript, and you can copy the link from the meeting page to share it with people who were not in the call. The page, its transcript and its link expire 3 days after the call ends. During the call the agent answers in the chat at once when it has something to say (a short quick reply written by DeepSeek V4.1 Flash, as described in Privacy), reacts to the line it heard with an emoji on its video tile, and, when something needs looking into, looks it up in the background in the meeting’s material and on the web (and, if you chose “everything” in its settings, in its memory of you, your files and the connected services you leave on for that call on the meeting’s page, where anything it would change or send waits for your approval), follows up in the chat with what it found, keeps on the page what is worth keeping for the meeting, and marks where people disagree (who holds which position, and the source behind each) and what is decided (with its own note on risks or conflicts with the meeting’s sources). The bot never speaks. Its video tile shows a page of what it is doing, what it has added and what is still to settle, drawn from the context page, and it writes only in the meeting chat, which every participant can read. CREAO does not record audio or video. Recall.ai processes the call’s audio, chat messages and participants’ display names in the United States, and keeps no recording and no transcript after the call. It streams the call’s audio to ElevenLabs, which transcribes it in whatever language is spoken, under the ElevenLabs terms described in Privacy. CREAO holds the transcript lines (speakers’ display names and what was said or typed, including by people who are not CREAO users) for the length of the call, at most 8 hours. When the call ends, a copy is kept with the page until it expires, and the transcript, headed by the page’s link and the names of the people seen in the call, goes to your personal agent, which writes a summary in your own conversation with it; that copy is removed 30 days after it is delivered, and the summary follows normal conversation retention. Only you see the summary. With the “everything” setting, if your Gmail or Outlook is among the services left on for the call and the attendees’ email addresses are known, the agent drafts one recap email to them, which waits for your approval on the meeting page and is never sent without it. To remove the bot, anyone who can remove participants from the call can remove it, and you can take it out yourself. It leaves by itself about a minute after you leave the call and do not come back, about 30 seconds after everyone else has left, when the call ends, after 10 minutes waiting to be admitted, or after 4 hours at the latest. See Privacy for what is sent to AI models before and during a call and for calendar data.
Being removed revokes your access to the team workspace, but does not automatically delete your files there. Team deletion also does not automatically erase each member’s private workspace files. There is currently no automatic deletion deadline for these files after departure. To request erasure, contact privacy@creao.ai. See Data Retention for details.
CREAO does not train its own models on your conversations, files, or data. Anthropic, OpenAI, Google, MiniMax, Z.AI, Moonshot AI, Thinking Machines, DeepSeek (open weights), Fireworks AI, Meta (Muse Wonder, Muse Ember and Muse Spark models via the Meta Model API), and BytePlus are accessed via API agreements or provider API terms that prohibit use of your data for model training. MiniMax M3 conversation requests are served by Fireworks AI under CREAO’s Fireworks account (Fireworks’ API terms prohibit training use) — if Fireworks returns a capacity error, the request is automatically retried via OpenRouter and served by DeepInfra, Novita, or Together AI under OpenRouter’s API terms, which prohibit training use — while MiniMax’s own API serves MiniMax M3.1 Flash conversation requests directly (including attached images) and CREAO’s internal response-quality evaluation, under MiniMax’s API terms, which prohibit training use; Moonshot AI, Thinking Machines’ Inkling and Qwen3.8 Omni Flash (served by Alibaba Cloud, the only host OpenRouter offers for it, so its conversation content and attached images reach OpenRouter and Alibaba Cloud) are reached via OpenRouter, whose API terms prohibit training use (Kimi K3 is served by Fireworks AI under CREAO’s account, and images attached to a Kimi K3 conversation reach OpenRouter and Fireworks; Kimi K3.1 Pro is served by Moonshot AI itself, the only host OpenRouter offers for it, so its conversation content and attached images reach OpenRouter and Moonshot AI), and GLM 5.2 / Moonshot AI requests may be served by Fireworks AI under CREAO’s Fireworks account — Fireworks’ API terms also prohibit training use. GLM 5.3 and GLM 5.3 Flash are the exceptions among the Z.AI models: they are sent directly to Z.AI under CREAO’s own Z.AI API agreement, which prohibits training use, so by default their conversation content reaches Z.AI and no other company — not even a relay. CREAO can also route GLM 5.3 (not GLM 5.3 Flash) to Fluxion, which serves it from its Asia region under CREAO’s Fluxion account; this is not the default route, and Fluxion is not covered by this no-training statement until CREAO confirms its data-use terms. (GLM 5.2 is still reached via OpenRouter. CREAO can move the GLM 5.3 models onto that path as a fallback, in which case OpenRouter relays the request and Z.AI still performs the inference.) Muse Spark 1.3 and 1.2 go directly to Meta by default, and reach OpenRouter only if CREAO moves them onto that path as a fallback, in which case OpenRouter relays the request and Meta still performs the inference, being the only host OpenRouter offers for those models (Muse Spark 1.1 has no such fallback). Muse Wonder 1.0 and Muse Ember 1.0 take the same path as Muse Spark 1.3, with the same OpenRouter fallback. DeepSeek V4.1 Flash is served by default through OpenRouter, preferring Fireworks AI under CREAO’s Fireworks account. Conversation messages, context, and attached images reach OpenRouter and the serving inference host; OpenRouter may use another host if Fireworks is unavailable. DeepSeek’s own API is excluded from Flash routing. These paths prohibit training use under their API terms. CREAO can also route Flash to Fluxion, which serves it from its Asia region under CREAO’s Fluxion account; this is not the default route, and when CREAO uses it, conversation messages, context, and attached images reach Fluxion instead of OpenRouter. The no-training statement above does not cover Fluxion until CREAO confirms Fluxion’s data-use terms. Selecting Flash does not keep content exclusively on CREAO-managed systems; the self-hosted V4 checkpoint is a separate, explicitly selected legacy route. DeepSeek V4 Pro uses a different serving provider: it is NOT self-hosted — every conversation request for it is relayed via OpenRouter and served by DeepSeek’s own API by preference (GMICloud as a fallback), so conversation content for that model IS sent to DeepSeek, under DeepSeek’s and OpenRouter’s API terms, which prohibit training use. Providers may retain data briefly for abuse monitoring and safety per their terms, but not for training on those paths. Fugu Ultra via Sakana AI is a qualified exception while governed by Sakana’s public Fugu terms: submitted content may be used for Sakana training or improvement unless an opt-out or separate no-training/DPA arrangement is in place. Team-room routing sends a room post that @mentions no agent, with recent room context, to TypeSafe’s Jev decision model via OpenRouter, which returns scores only; TypeSafe’s published privacy policy says it does not train or fine-tune models on that input, keeps it as long as reasonably necessary to provide its service rather than for a fixed period, and processes it in the United States.Music prompts (including lyrics) and generation settings are sent to ElevenLabs when music generation is enabled; selected Scribe transcription requests also send audio to ElevenLabs. Its training and retention terms differ from the API commitments above. See the ElevenLabs disclosure in Privacy; CREAO does not promise zero retention for music.
CREAO stores conversation files on cloud infrastructure in the United States. Data is encrypted at rest using AES-256 encryption and in transit using TLS 1.2+. Each chat thread runs in an isolated sandbox that is destroyed after the session ends. Team-room routing requests sent to TypeSafe’s Jev decision model are processed by TypeSafe in the United States.Team-room human requests also store the agent’s ask and your answer. Requests expire after 14 days and are deleted 90 days after closure; account erasure removes requests you own or settled and your recipient records. See Data Retention.ElevenLabs uses US storage by default, but its international affiliates and subprocessors may process content elsewhere. Provider retention and transfer terms are described in Privacy.
CREAO estimates your approximate location from your IP address at login. A country code is used for analytics, abuse detection, and billing routing, and CREAO may also derive a coarse city/region estimate, used for regional analytics and to segment location-relevant CREAO marketing and lifecycle emails by area. This estimate is approximate — it reflects your network, not a precise position, and can be wrong (VPNs, mobile carriers). CREAO does not collect GPS or precise-location data, and only the coarse city/region estimate — never your raw IP — is shared with our marketing-automation provider (customer.io) for this. Because CREAO does not yet operate a per-user consent mechanism for location-based marketing, users in the EU/EEA/UK are excluded from it entirely. You can opt out of marketing emails at any time via the unsubscribe link or your notification settings. See Privacy for details.
Yes. You can delete individual conversations, files, and memory entries at any time from the respective pages. Where memory stores are enabled, ask the agent to forget a memory it saved; deleting an agent deletes the memories it kept. To delete your CREAO account and its associated data (agent.creao.ai and CLI), go to your account settings, where you can optionally tell us why you’re leaving. Account deletion is processed within 30 days. A Developer Platform account created at developer.creao.ai is not removed by that request; email privacy@creao.ai to erase it. A limited audit record of the deletion request itself (such as request metadata, an email snapshot, any reason you shared, and any administrator handling notes) may be retained afterward where required for legal, compliance, and security purposes, as described in the Privacy policy.If agent alignment is enabled in the preview, the companion can review its own conversations with you, your feedback, its actions and the room replies to its posts that you can access with its model provider, then propose learned rules from your alignment-card answers. A separate confirmation shows the exact replacement text before saving. Rules apply across the companion’s later chats and rooms and may contain names or paraphrases; room headings are not access restrictions. Your room replies may also reach another member’s companion model for their alignment review. Learned rules have no automatic expiry. Agent or team deletion does not erase these files. Account hard deletion removes current files. Historical versions may remain after replacement or deletion until a separate verified purge; do not rely on automatic expiry during this preview. Contact privacy@creao.ai for historical erasure. Deleting your account does not remove names or paraphrases already saved in another person’s learned rules. Contact privacy@creao.ai to request erasure of those copies. Alignment cards save quoted evidence, including other members’ room replies, in the owner’s home conversation; those copies follow normal conversation retention and existing analytics warehouse handling. Deleting the source member does not erase copies in another owner’s conversation; request their erasure at privacy@creao.ai.Every turn in your personal companion’s home conversation is also stored as a run record with the first 500 characters of your message, whether or not alignment is enabled. These records follow existing run/session retention and analytics warehouse replication.If the personal companion preview is enabled, deleting its team disables the companion and erases its encrypted soul notes and remembered facts, but retains the rest of its configuration (including onboarding profile, home-conversation identifier, daily check-in settings, and whether you used CREAO before it and closed its walkthrough) in the database and analytics warehouse, with no automatic hard-purge deadline. Team deletion does not erase this data; home and delegated conversations follow normal conversation retention. A remembered fact you correct keeps its earlier text for 90 days; facts kept before they moved to its memory store are no longer shown in its menu and stay stored until its team or your account is deleted. Submitting an account-deletion request does not itself erase companion data. The administrator-performed account hard delete removes user-owned companion records, soul notes, remembered facts, daily check-in settings, suggested changes to your agents (which team deletion keeps with the agent’s record) and conversations, and removes your agents from other teams’ rooms where it can (a room it could not clean is recorded), whose messages stay as the rooms’ history; warehouse replicas follow source deletion. Contact privacy@creao.ai for an erasure request or about anything that remains.
CREAO is designed with GDPR and CCPA principles in mind. You have the right to access, export, and delete your data. We collect only the minimum data necessary to provide the service and do not sell your personal information. Enterprise customers can request a Data Processing Agreement (DPA). See the Privacy page for full details.
Secrets are encrypted at rest and managed via a dedicated secrets store. They are injected into your sandbox as environment variables at runtime and are never logged, displayed in the UI after creation, or stored in plain text. For S2 debot trading, the participant’s own debot developer API credentials (DEBOT_API_KEY and DEBOT_API_SECRET) are treated as user secrets and are injected only for the participant’s Agent runtime when it executes a live Robinhood Chain (EVM) action through debot’s custodial wallet API. No wallet private key is collected or stored — debot signs transactions server-side. In a team room, room owners and admins can add room secrets: room members from the team see their names (guests don’t), never their values, and every agent running in that room receives them, so anyone who can message those agents, guests included, can make them use one.
Only you can see your conversations by default. If you use the Share feature, a read-only public link is generated — anyone with the link can view that specific thread and download any files or artifacts within it. Shared links can be revoked at any time. Team members in a shared workspace can see workspace-scoped agents and files, but not your private conversations.

​
Files and Data

HTML pages, PDFs, SVGs, spreadsheets (XLSX, CSV), images, code files, documents, and more. Generated files appear in the artifact preview panel during chat and are saved to your Files page automatically.
Yes. Drag and drop files into the chat input or click the attachment icon. Supported types include images (PNG, JPG, GIF, WebP), documents (PDF, TXT, CSV), and code files (HTML, CSS, JS, Python, etc.).