Google Chat Chatbots: Setup, Permissions, and Follow-Up Work

A Google Chat chatbot is a workspace app that responds to @mentions and slash commands inside Chat spaces and direct messages. It can answer questions, trigger workflows, and connect to external services. Tools like fenbs record the agreed decisions and follow-up work a chatbot session produces — in a place where a separately connected AI assistant can read and act on them, independent of the chat conversation itself.

Setting up a Google Chat chatbot involves a few interconnected decisions: where the bot's code runs, what events it receives, and who can add it to spaces. Getting those choices clear upfront saves time when something does not behave as expected.

What a Google Chat chatbot can do

A Google Chat app can receive and respond to:

• Direct messages: users can send a message directly to the bot as a DM.
• @mentions in spaces: the bot receives a message event when someone @mentions it.
• Slash commands: users type /command and the bot receives a slash command event with the command name and any arguments.
• Card interactions: the bot sends a card (a structured message with buttons and fields) and receives an event when the user interacts with it.

The bot does not receive every message in a space — only ones directed at it. This is intentional. A bot that reads every message in a large space raises privacy concerns and creates a lot of noise to filter.

How to add a chatbot to Google Chat

1. Create a Google Cloud project and enable the Google Chat API at console.cloud.google.com↗ (opens in a new tab).
2. In the Chat API configuration, set the bot's display name, description, and avatar. Choose the connection type: HTTP (you host the webhook endpoint) or Apps Script (Google hosts it).
3. Write the event handler code. For HTTP bots, this is a server that receives POST requests from Google Chat. For Apps Script bots, this is a function in an Apps Script project.
4. Under "App home", configure which event types the bot should receive: direct messages, @mentions, slash commands, or all of the above.
5. For internal testing, add the bot to a test space by @mentioning it or through the Chat API configuration.

Check Google's Chat API documentation↗ (opens in a new tab) for the current setup steps. The details change between API versions.

Workspace permissions for chatbots

Google Workspace admins control which Chat apps users can install. If your organization restricts app installation, users cannot add the bot to spaces without admin approval. You may need to submit the app for admin review or add it through the Admin console yourself.

For apps that access sensitive data — Drive files, calendar events, email — the OAuth scopes you request go through a verification review if the app is published externally. For internal apps used only within your organization, verification is not required.

Testing your chatbot before team rollout

Test in a private space before adding the bot to team spaces. Send it the expected inputs, edge cases, and failure cases deliberately. Check what happens when it receives a command it does not recognize, an input that is too long, or a request it should decline.

Share the test space with one or two teammates. Their questions will be more realistic than your own. Pay attention to where the bot gives a plausible but incorrect answer — those cases usually point to a system prompt or instruction that is too vague.

Following up on bot interactions

Chat conversations move fast and older messages are hard to find. Decisions made in a chatbot interaction — a question answered, a request approved, a next step confirmed — do not automatically become tracked work. They need to be pulled out of chat and put somewhere they can be acted on and reviewed.

This applies both to decisions the bot helped make and to tasks that a bot interaction generated. If the bot confirms that a design change is approved, that approval should be recorded where the design work is tracked — not just in a chat message. See Google Tasks for integrating task creation directly from Chat, and Google AI tools for the broader ecosystem.

fenbs keeps Decisions and rules and task tracking in a shared record outside Chat. A separately connected AI assistant reads those before starting work — so the chatbot interaction's outcome is available to the next person who picks up the project.

How to record chatbot decisions and follow-up work with fenbs

1. Create a free account and add a project for the work your chatbot supports.
2. When a chatbot interaction confirms a decision — an approved change, a chosen approach, an agreed next step — add it as a Decision in fenbs with the date.
3. When a chatbot interaction generates a task — a follow-up needed, a check to run, a change to make — add it as a task in fenbs.
4. Connect your AI assistant over MCP from Integrations. Ask it to read the project context before helping with the work the chatbot interaction generated.
5. Review the Decisions and rules periodically. Remove ones that no longer apply and update any that have changed.

fenbs does not connect natively to Google Chat. It works alongside it: a chatbot handles interactions inside Chat, and a separately connected AI assistant reads the project record in fenbs before taking action. The chatbot talks to your team; the assistant works on the outcomes of those conversations.