How can we help?

MeetCrew is an AI meeting participant that joins Microsoft Teams, Zoom, Webex, and Google Meet as a named, disclosed teammate — an avatar, a voice, and persistent cross-meeting memory that fronts the chat agent you already run.

In the admin console choose Create deployment, give it a name and wake word, pick a character and voice, connect your backend, and authorize the meetings it may join.

Microsoft Teams, Zoom, Webex, and Google Meet. Available in the US and Canada in v1.

A teammate joins only meetings a tenant admin authorized, and only in supported regions (the US and Canada). Check both, then contact support if it still won't join.

Yes. A teammate discloses itself three ways — a named participant with an AI badge, an on-screen banner in every frame, and an audible self-introduction — and the disclosure can't be turned off by end users.

Say '[Name], please leave' or 'don't remember this conversation.' It's honored immediately and logged.

Topics, decisions, action items, open questions, and commitments — carried from one meeting to the next, scoped by your admins.

No. MeetCrew never creates or stores voiceprints or any voice-derived biometric. Cross-meeting recognition uses Microsoft Entra platform identity, never voice.

Your meeting memory is stored only in Microsoft Azure, in the US East region, where every MeetCrew tenant is hosted today, and it stays in the United States: any backup or recovery copies Azure keeps, including in its paired region, are in the United States too. Meeting content leaves Azure in three places. Meeting audio is carried by our meeting-connectivity provider and streamed to Deepgram, our speech-to-text provider, which transcribes it on its own hosted service outside Azure, not pinned to a region; that audio is opted out of Deepgram's model training and is not kept after processing. When a teammate's animated avatar is on, its own synthesized speech is carried by Cloudflare to a GPU server MeetCrew operates itself in the United States, which animates the avatar's face. And when a teammate answers, the request goes to your own backend agent, wherever you host it.

No. The capture layer runs Zero Data Retention by default — the bot vendor stores no recording at any point, and access is real-time stream only. Your durable meeting memory exists only in your MeetCrew tenant store, under your retention policy.

Never. Customer meeting content is not used to train any model, ours or anyone else's.

Email privacy@meetcrew.ai. Separately, meeting transcripts and the memory extracted from them are deleted automatically after 24 months.

Yes. On an admin-mediated data-subject deletion request, the individual is removed from the memory teammates and people can query and search — their name is regenerated out of the structured records and the semantic index, not merely hidden — and a durable marker keeps a later re-extraction from reintroducing it. The raw source transcript is kept as a non-queryable record until its retention period ends; v1 has no option to delete it on request.

In the deployment's Configuration, bind Copilot Studio, Microsoft Foundry Agent Service, an OpenAI-compatible endpoint, or a custom webhook. You enter its credential once; MeetCrew keeps it in a dedicated Azure Key Vault and never shows it again.

Tenant administrators authorize meetings and decide which teammates exist and who can configure them.

Per named teammate, not per seat. Each plan includes monthly agent-hours — one teammate in one meeting for one hour; see the pricing page for details.

Yes — each teammate is a deployment you manage in the admin console. Adding one may move you to the next plan.

Copilot Studio, Microsoft Foundry Agent Service, any OpenAI-compatible endpoint, or a custom webhook.

They let your own agent read structured meeting memory — decisions, open questions, action items, statements, and topics, with Entra-based attribution and timestamps — and act on the workflow records: close an action item, resolve an open question, or update a decision. There's no delete, and nothing can alter a transcript, a statement, provenance, or the audit log.

Only an agent your organization authorizes. Access runs under a Microsoft Entra service principal your admin grants in the console, and MeetCrew authenticates every call against that identity — there's no shared key or MeetCrew-side access to your memory.

No. Every call is scoped to a single tenant and enforced on our servers, not by the client — a caller can't read or change another tenant's memory, including by guessing record IDs.

No. Attribution comes from Microsoft Entra platform identity, not voice. The interface exposes no voiceprints, no speaker audio, and no biometric data, and the memory is stored only in Azure.

By default each teammate's memory is its own: the teammate in a meeting, and your agent through a Memory API grant for that teammate, reach only its records. A Tenant Administrator can share one teammate's memory with another teammate in the same tenant, from the Sharing tab of the teammate whose memory it is. Sharing runs one way and gives read access only: sharing Jim's memory with Todd lets Todd read Jim's records but not change them, and gives Jim nothing of Todd's.

Yes. Every read and every action over the MCP server and API is logged and attributable to the Microsoft Entra identity that made it, so you can see who did what and when. The two are retained differently, which matters if you are answering a question long after the fact. Write actions — creating an action item, resolving an open question, updating a decision — are recorded as durable audit events in the tamper-evident audit archive and retained for seven years. Reads are recorded in operational logs and retained for 90 days. Both name the identity, the deployment and the operation, so a read is fully attributable inside that window and is not attributable after it.