This project demonstrates how to connect a Google Dialogflow CX agent to Azure Bot Services, and then integrate that bot into Microsoft Teams as a personal assistant that can relay messages into a Teams channel.
This is a rare cross-cloud setup between Google Cloud and Azure, and due to the lack of official documentation, I had to reverse-engineer much of the process through trial, error, and deep debugging.
There was no reliable end-to-end documentation on how to:
- Route Dialogflow CX responses through Azure Bot Services.
- Package that bot for use inside Microsoft Teams.
- Relay conversations into Teams channels with the correct permissions and scopes.
Even Microsoft’s advanced support didn’t have a clear path forward.
To solve this, I:
- Connected Dialogflow CX to a GCP-hosted webhook.
- Registered an Azure Bot to relay those messages.
- Built a Teams-compatible bot app using the Teams Developer Portal.
- Published it to Teams for personal chat use and team channel integration.
- Configure intents, flows, and fulfillment in Google Cloud.
- Enable webhook responses and test within Dialogflow.
- Deploy a Cloud Function or Cloud Run instance that acts as a proxy.
- Accept
POSTrequests from Azure and route them to Dialogflow CX. - Return CX responses formatted for Microsoft Bot Framework.
- Create an Azure Bot resource.
- Set the messaging endpoint to your GCP webhook.
- Save your Microsoft App ID and secret for later use.
This was the most complex part. Here's what worked for me:
- Go to Teams Developer Portal and create a new app.
- Use the same Microsoft App ID from your Azure Bot registration.
- Add app icons and basic branding.
- Under App Features, define this app as a Bot.
- Use the Azure App ID for the bot.
- Scope the bot to Personal so it can respond 1:1.
- Add domains required for communication:
*.botframework.com- Your production hosting domain (if needed)
- Use the built-in validator to check for issues.
- Publish the app internally to your org.
- Make sure a Teams admin approves the app and sets proper access.
- 🔄 Permissions can take several minutes to propagate—be patient.
- ✅ An admin must approve app permissions or it won’t function.
- 🧪 Use the “Test in Web Chat” feature under your Azure Bot to validate CX responses before installing in Teams.
- 🔐 Ensure the app isn’t blocked by org policy.
- 🔁 Set Teams scopes correctly in the Admin Center to allow channel relays.
- Users can message the bot in Teams, which relays input to Dialogflow CX.
- CX generates a response which is returned through Azure Bot to Teams.
- All queries and responses are captured in a Teams channel for visibility.
- Multi-cloud integration is very real—and very painful without docs.
- Testing in isolated layers (GCP → Azure → Teams) helped massively.
- Teams bots require very precise scope