The difference between an AI chatbot and an AI operator is integrations. Before connecting to real tools, an agent can only talk about work. After integrations, it can actually do work.

This guide walks through connecting your OpenClaw agent to the three tools that form the backbone of most startup operations: Gmail, Slack, and Linear. Follow along and you'll have a fully connected AI agent in about an hour. If you want help scoping which integrations matter most for your business, start with an AI Strategy Audit.

Why These Three?

Gmail, Slack, and Linear cover the three core loops of a startup:

Gmail — external communication. Customers, investors, partners, vendors.
Slack — internal communication. Team coordination, quick decisions, status updates.
Linear — work management. What needs to be built, who's building it, what's blocked.

When your AI agent can read and act across all three, it stops being a tool you use and starts being a teammate you work with.

Prerequisites

Before you start, you'll need:

An OpenClaw account — sign up at torontoaiconsulting.com
Admin access to your Gmail, Slack workspace, and Linear team
About 60 minutes of uninterrupted setup time
A willingness to give an AI access to your communication channels (this is the hard part, and we'll talk about it)

Part 1: Connecting Gmail

Gmail integration gives your agent the ability to read incoming emails, draft responses, and send messages on your behalf. This is where most people get nervous — and that's reasonable.

Step 1: Google Workspace Setup

OpenClaw uses OAuth 2.0 to connect to Gmail. This means your agent never sees your password. Instead, Google gives OpenClaw a token with specific permissions that you can revoke at any time.

Navigate to the OpenClaw dashboard and go to Settings → Integrations → Gmail. Click "Connect Google Account."

You'll see the standard Google consent screen. The permissions requested are:

Read email — so your agent can check your inbox
Send email — so your agent can send on your behalf
Manage labels — so your agent can organize your inbox

Step 2: Configure Access Scope

This is important. OpenClaw lets you configure exactly what your agent can do with Gmail:

# Example Gmail config in OpenClaw
gmail:
  read: true
  send: true
  send_requires_approval: true  # Agent drafts, you approve
  auto_label: true
  accounts:
    - primary@yourdomain.com

We recommend starting with send_requires_approval: true. This means the agent drafts emails and shows them to you before sending. After you've reviewed a dozen drafts and trust the quality, you can flip it to false.

Step 3: Test the Connection

Ask your agent to check your inbox:

"Hey, check my email — anything urgent?"

Your agent should be able to pull recent messages, summarize them, and flag anything that needs attention. If this works, Gmail is connected.

What Your Agent Can Do with Gmail

Here's what a typical morning workflow looks like with Gmail access:

Scan the inbox for anything that arrived overnight
Categorize messages: urgent / needs response / informational / spam
Draft responses for anything straightforward
Flag anything that needs your judgment
Create Linear issues for any customer bugs or feature requests mentioned in email

That last step — turning emails into Linear issues — is where the integrations start compounding. For more on building an AI email assistant, we have a dedicated guide.

Part 2: Connecting Slack

Slack integration is the real-time nervous system. Gmail is async; Slack is live.

Step 1: Create a Slack App

In your Slack workspace, go to api.slack.com/apps and create a new app. OpenClaw provides a manifest that pre-configures the right permissions:

{
  "display_information": {
    "name": "OpenClaw Agent",
    "description": "Your AI teammate"
  },
  "features": {
    "bot_user": {
      "display_name": "openclaw-agent",
      "always_online": true
    }
  },
  "oauth_config": {
    "scopes": {
      "bot": [
        "channels:history",
        "channels:read",
        "chat:write",
        "im:history",
        "im:read",
        "im:write",
        "users:read"
      ]
    }
  }
}

Step 2: Install and Configure Socket Mode

OpenClaw uses Socket Mode rather than webhook URLs. This means:

No public endpoints needed
Lower latency
More reliable connection

Enable Socket Mode in your app settings and generate an App-Level Token. Paste this into your OpenClaw config.

Step 3: Set Channel Policies

Define where your agent can and can't participate:

DM policy: pairing — the agent responds in direct messages
Channel policy: allowlist — specify which channels the agent can see and respond in

Start with just DMs and one or two channels. Expand as you get comfortable.

Step 4: Test

Send a DM to your OpenClaw bot in Slack:

"What's on my calendar today?"

If it responds with your schedule (assuming calendar is connected), Slack is working. For a deeper dive, see our full guide on using OpenClaw with Slack.

What Your Agent Can Do with Slack

In Slack, your agent operates as a team member rather than a personal assistant:

Answer questions about the codebase when developers ask
Post daily standups summarizing completed work
Respond to mentions with context from Linear or email
Share relevant metrics when someone asks "how are we doing on X?"

The key is knowing when to speak and when to stay quiet. Nobody wants a bot that responds to every message. A well-configured agent follows the same rule a thoughtful human colleague would: contribute when you have something useful to add.

Part 3: Connecting Linear

Linear is where work gets tracked. Connecting your agent to Linear means it can create issues, update statuses, add comments, and pull project data.

Step 1: Generate an API Key

In Linear, go to Settings → API → Personal API Keys and create a new key. Give it a descriptive name like "OpenClaw Agent."

Step 2: Add to OpenClaw Config

linear:
  api_key: lin_api_xxxxxxxxxxxxx
  default_team: "Engineering"
  can_create_issues: true
  can_update_issues: true
  can_comment: true

Step 3: Configure Teams and Workflows

OpenClaw needs to understand your Linear setup — teams, workflow states, labels, and projects. The first time it connects, it maps these automatically. But it's worth reviewing:

Teams: Engineering (ENG), Marketing (MKT)
States: Backlog → Todo → In Progress → In Review → Done
Labels: bug, feature, improvement, urgent

Step 4: Test

"Create a Linear issue: Fix the login page timeout bug, assign to Engineering, priority high"

Your agent should create the issue with the right team, state, and priority. For a full walkthrough of AI-powered project management with Linear, check out the dedicated guide.

What Your Agent Can Do with Linear

Linear is probably where an OpenClaw agent adds the most value:

Triage incoming issues — when bugs come in from email or Slack, the agent creates properly formatted Linear issues with reproduction steps
Update issue statuses — when a coding task is complete, the agent moves the issue to "In Review"
Add context — the agent comments on issues with relevant information from email threads or Slack conversations
Generate reports — weekly summaries of what shipped, what's in progress, and what's blocked

The magic happens when all three integrations work together. A customer emails about a bug → the agent creates a Linear issue → posts in the engineering Slack channel → follows up when the fix ships. That entire loop used to require a human coordinator.

The Compound Effect

Each integration is useful on its own. Together, they're transformative. Here's a real example:

9:14 AM — Customer email arrives reporting that exports are timing out
9:15 AM — The agent reads the email and classifies it as a bug report
9:16 AM — A Linear issue is created with the error details and a link to the customer's account
9:17 AM — A message posts in #engineering Slack: "New P1 bug — exports timing out for enterprise accounts. Linear: ENG-342"
9:18 AM — A draft reply is prepared for the customer: "We've identified the issue and our engineering team is investigating. I'll follow up when we have a fix."
9:19 AM — You review and approve the customer reply

Six minutes. No human context-switching. No "let me check with the team." No issue falling through the cracks.

Security Considerations

Giving an AI agent access to your email, chat, and project management means:

The agent can read sensitive communications
The agent can send messages on your behalf
The agent has context about your entire business

OpenClaw mitigates this with:

OAuth tokens (not passwords) for all connections
Granular permissions (read-only vs. read-write)
Approval workflows (agent drafts, human approves)
Audit logs (every action is logged)
Revocable access (disconnect any integration instantly)

But ultimately, you're making a trust decision. Start with read-only access. Graduate to drafts-with-approval. Only go fully autonomous when you're confident. For a deeper look at security practices, see our guide on OpenClaw security skills.

Common Setup Issues

Gmail: "Access Denied" error

Make sure you're using a Google Workspace account (not a personal Gmail)
Check that your admin has allowed third-party app access

Slack: Bot not responding

Verify Socket Mode is enabled
Check that the bot is invited to the channels you're testing in
Confirm the DM policy is set to "pairing"

Linear: "Unauthorized" error

API keys can expire — regenerate if needed
Make sure the key has the right scopes

What's Next

Once you have Gmail, Slack, and Linear connected, you've built the foundation. From here, you can add:

GitHub for code-aware development workflows
Calendar integration for scheduling
PostBridge for social media automation
DataforSEO for automated ranking reports

Each new integration adds another dimension to what your agent can do. But these three — email, chat, work management — are the core. Get these right and everything else is incremental. If you need help designing a custom integration stack, Toronto AI Consulting offers MCP & API integration services to get you set up fast.