Microsoft Teams app
Govern your AI agents from Microsoft Teams.
High-risk alerts and dual-control Approve or Deny on held tool calls, right inside a Teams chat or channel. Bot Framework verified callbacks, a one-time link code, and a tamper-evident audit trail.
Adaptive Card approvals · dual control · decision in under 100ms
What you get
The moment that needs a human, brought into Teams.
AxioRank scores every tool call, enforces policy, and keeps a provable audit trail. The Teams app surfaces what a person actually has to act on, and lets them act without leaving the chat.
High-risk alerts
When an agent does something risky, an Adaptive Card lands in the chat you choose, with the agent, tool, risk score, and decision.
Approve or deny, in Teams
Held tool calls arrive with Approve and Deny actions. Two-person holds need two distinct Teams users, the same dual-control path as the dashboard.
Provable trail
Every decision is written to a tamper-evident audit log you can verify offline, so an approval in Teams is real evidence, not a screenshot.
Setup
Linked in about a minute.
Add the app to a chat, paste the link code from the Channels page, and you are done. No routing key, no webhook URL to copy.
1. Add the app
Install the AxioRank app into a Teams chat or channel from your organization's app catalog. The bot captures the conversation it was added to.
2. Paste the link code
Copy the short-lived code from the AxioRank Channels page into that chat. It binds the conversation to your workspace, with no token to wire up by hand.
3. Approve in Teams
Held calls show up with Approve and Deny. Resolve them in Teams or the dashboard; either way the decision is recorded once and audited.
Trust
Least privilege by design.
A security control should not be the thing that widens your attack surface. The bot acts only on the conversation it was added to, verifies every callback, and reads no messages.
No message reading
No history or content permissions. The bot never sees chat content.
Verified callbacks
Every Approve or Deny is checked against a Bot Framework JWT, the bot's app id, and the conversation service URL before it acts.
Deliberate linking
A short-lived, signed link code binds a conversation to exactly one workspace, so a chat is never bound by guesswork.
Questions
Good to know before you install.
The security questions an admin asks first.
Does AxioRank read our Teams messages?+
No. The bot only acts on the conversation it was explicitly added to, and only on its own Adaptive Card actions and the one-time link code you paste. It does not read channel history and requests no message-content permissions.
How does dual control work in Teams?+
Approve and Deny actions run through the same approval engine as the dashboard and signed email links. When a policy requires two approvers, a hold needs two distinct Teams users to clear it. Every action callback is verified with a Bot Framework JWT, checked against the bot's app id and the conversation's service URL, before any vote is recorded.
What gets stored when we connect?+
The conversation reference for the chat you linked, so AxioRank can post alerts and approval prompts there. No message content is stored. The bot credentials live in deployment configuration, not on the channel row.
Why a link code instead of a click-through install?+
A Teams bot install does not carry your AxioRank workspace identity the way a web OAuth flow does. The short-lived, signed link code proves which workspace a conversation belongs to, so a chat is bound to the right tenant deliberately rather than by guesswork.
Which plans include the Teams app?+
Teams notifications and in-Teams approvals are available on every plan. Higher tiers add more governance surface, longer audit retention, and the AI assessment models. See pricing for the full breakdown.
Bring AI agent governance into Microsoft Teams
Add the app to a chat, paste the link code, and start approving held tool calls where your team already works.