Microsoft Teams

Once your organization's Teams bot is provisioned, you can chat with Strata directly inside a Microsoft Teams conversation — no browser, no extra sign-in.

[User] Chatting with the Strata bot in Teams

After your IT admin installs the Strata Teams app into your tenant, find the Strata bot in your Teams chat list and send a message. The bot reuses the same user account, settings, and data your web app is bound to (resolved through your Microsoft AAD object ID), so anything you've already configured — connected SQL Server, OneDrive, Outlook, AI model, mode, permission tier — applies in Teams too.

If you've never signed in to Strata on the web before, the bot replies with:

Please sign in to Strata at the web app first to set up your account.

Go to the web app, complete the Microsoft SSO sign-in, then come back to Teams and try again.

[User] Commands

The Teams bot understands a small set of typed commands plus natural-language questions:

Anything else you type is treated as a natural-language question and sent to the AI exactly as if you'd typed it in the web app's chat.

The model IDs the bot recognizes come from your organization's Model Catalog — the same catalog the web app's model picker draws from. The Settings card (type settings) narrows its picker to the models your org, role, and per-user access allow, so a model you can't use never appears there. The typed /model command recognizes any catalog model ID; if you set one that policy doesn't permit you to use, the bot enforces governance by clamping back to your first allowed model on your next message (see the note below). If you send /model with an ID that isn't in the catalog at all, the bot replies with the list of model IDs it recognizes.

[User] Selecting model, mode, and permission per user

Open the Settings card (type settings) to pick:

• AI Model — the picker lists the models from your organization's Model Catalog that you're allowed to use. Strata runs a central Model Catalog (curated by Kronisys, served through Microsoft Azure AI Foundry) spanning providers including OpenAI, Anthropic, xAI, and more; your admin enables a subset for the org, and your role and per-user access can narrow it further. A disabled model never appears as a choice. See Models for the full picture.
• Response Mode — Auto (adapts to each question), Technical (code and detail), or Simple (plain language only).
• Permission Level — Viewer (read-only queries), Editor (read + write), or Admin (full access including DDL).

Click Save Settings. Choices persist to the same user_settings row your web app reads, so changing in Teams updates your defaults everywhere.

Note

The bot enforces model governance exactly like the web app. If your selected model is later disabled for your org, the bot falls back to your first allowed model on your next message — your choice isn't honored if policy no longer permits it.

[User + Admin] What works in Teams vs. the web app

Teams covers the core conversational surface: natural-language questions, SQL execution, Excel and PDF generation (delivered as Adaptive Cards with download links), and conversation history. The web app remains the place for:

• Connecting new SQL Server databases, OneDrive, or Outlook for the first time. Teams reads the connections, but the credential / OAuth flows happen on the web.
• The full artifact panel, dashboards, voice mode, and incognito chat.
• All admin work — user management, roles, audit, agent governance, org settings.

OneDrive and Outlook sessions established on the web are automatically restored when the Teams bot resolves your session, so any extension you connect on the web is available to the bot.

Every AI call the bot makes routes through the same Azure AI Foundry the web app uses — your organization's own Foundry, in your own Azure tenant (see Bring your own Azure AI Foundry). Teams messages never touch any other engine. Usage limits and quotas also apply in Teams: a message that would exceed your org's budget is refused the same way it would be on the web.

[Admin] Setup overview

Teams setup is coordinated by Kronisys. Your IT team installs the Strata app into your Microsoft 365 tenant and governs it like any other Teams app. Regardless of the bot identity, every message is bound to your Strata account and routed through the same Azure AI Foundry the web app uses — your organization's own Foundry, in your own Azure tenant.

The steps the customer takes vs. the steps Kronisys takes are:

Expect the back-and-forth to take roughly one business day. The provisioning ticket includes your org label, org ID, the requesting admin's email, and a timestamp so the engineering team can wire everything to the right tenant.

[Admin] Managing the Teams extension's availability

The Teams extension is governed the same way as the other extensions:

• Settings → Extensions → Microsoft Teams → Manage → Organization settings — turn the availability switch off to remove teams from the org. The bot stops responding to org members and replies with a message telling them to ask an admin to re-enable Teams. Re-check to restore access. (Roles can narrow the extension set further than the org allows.)
• Removing the Teams extension does not uninstall the bot from users' Teams clients. To fully remove the integration, work with Kronisys to retire the bot registration.

What can go wrong

• "Please sign in to Strata at the web app first to set up your account." — your AAD object ID has no matching Strata user. Sign in once on the web (which provisions the user record) and retry in Teams.
• The bot doesn't respond at all — the bot may not be installed in your tenant yet, or your admin removed it. Check that you can see Strata under Apps > Manage your apps in Teams.
• db command shows "No Saved Databases" — you have no SQL Server connections saved. Add one in the web app under Settings > Extensions, then retry.
• The bot says "Database connection was lost..." — the in-memory connection pool was recycled (after 30 minutes of idle time, or after a service restart). Type db and reselect the connection to re-open the pool.
• A model in /model isn't recognized — the ID isn't in the Model Catalog at all. The bot replies with the list of model IDs it recognizes; pick one of those. If you set a model that is in the catalog but disabled for your org, role, or user, the bot accepts the command but clamps back to your first allowed model when you next send a question — so the effective model always honors policy. Use the Settings card to see only the models you're allowed to run.
• "Microsoft Teams is not enabled for your organization." — an admin removed teams from the org's allowed extensions (or your role excludes it). Ask your Strata admin to re-enable it under Settings → Extensions → Microsoft Teams → Manage.

Related

• Models — the Model Catalog and how model availability is governed.
• Organization settings — managing allowed extensions and the Model Catalog.
• Onboarding your organization — checking the Teams option during onboarding to kick off provisioning.
• Integrations overview — the extension model in one page.