• SEGMENTATION
  • EXPOSURE
  • VALIDATION
  • HUMAN

How to manage canary rollouts with your AI

Connect ConfigCat to your AI client to control feature flags through natural language. You can segment users and toggle features without touching the ConfigCat dashboard.

Ask AI about this page

Short answer

How can I use AI to handle canary rollouts?

Your AI uses ConfigCat tools to build specific user groups and change flag values on the fly. This lets you roll out updates to small percentages of users or specific cohorts instantly. You get granular control over feature exposure through your chat interface.

Rollout outcomes.

Where your control lands.

What happens when you direct your AI to manage your flags.

SEGMENTATION

Targeted user groups

The AI builds custom segments based on your criteria. You can isolate specific users for testing without manual configuration.

EXPOSURE

Controlled feature access

The AI updates flag values to turn features on or off for specific segments. This manages how much of your user base sees a change.

VALIDATION

Real-time flag checks

Your agent checks current flag values to confirm the rollout state. You'll know exactly which settings are active in your environment.

HUMAN

Manual oversight

You decide when to trigger a full rollout or a rollback. The AI handles the technical execution, but you keep the final say.

The workflow

What your AI does when a rollout starts.

The AI moves from identifying users to adjusting the actual feature toggles.

  1. Identify existing groups

    The AI looks at your current user segments to see who is already categorized.

    list_segments

  2. Create test cohorts

    The AI builds a new segment specifically for your canary test group.

    create_segment

  3. Verify current settings

    The AI checks the existing value of a feature flag to ensure a clean starting point.

    get_setting_value

  4. Execute the rollout

    The AI updates the flag value to enable the feature for your new segment.

    update_setting_value

Try it

Copy these to start.

Use these prompts to kick off your rollout management.

Starting points

Replace the bracketed text with your actual segment names or flag keys.

ConfigCat Connector

You're all set. Choose your MCP client and follow the setup instructions.

Connector linkhttps://edge.vinkius.com/vk_preview_P9Er6wmWUpWSD8h0DEvbhlBRaDOUdRbdP9tsPQXz/mcp

Claude Desktop

Follow the steps below to connect in seconds.

  1. 1In Claude Desktop, open Settings → Connectors.
  2. 2Click “Add custom connector” and paste the connector link above as the remote MCP server URL.
  3. 3Click Add and start a new chat — ConfigCat capabilities are ready to use.
Configuration · claude_desktop_config.jsonCopy
{
  "mcpServers": {
    "configcat-mcp": {
      "url": "https://edge.vinkius.com/vk_preview_P9Er6wmWUpWSD8h0DEvbhlBRaDOUdRbdP9tsPQXz/mcp"
    }
  }
}
Copy into chat04

  • Create a new segment in ConfigCat called [Beta Testers] for users in [New York].

  • What is the current value for the [new_ui_feature] flag?

  • Update the [canary_rollout] flag to [true] for the [Internal Employees] segment.

  • List all my current segments in ConfigCat so I can see who is available for testing.

  • Claude
  • ChatGPT
  • Cursor
  • VS Code
  • Windsurf
  • Claude Code
  • JetBrains
  • Cline

Start here

Connect ConfigCat once, then ask.

Link your account in one click. Your credentials stay encrypted, and you can immediately start managing flags through your AI client.

Connect ConfigCat to your AI

FAQ

How the ConfigCat MCP behaves

  • 01

    Can the AI delete my existing segments?

    No. The AI can only list, create, or update segments. It cannot delete any data.

  • 02

    Can the AI change my account billing settings?

    No. The AI only has access to segment and flag management tools.

  • 03

    Does the AI need to ask me before changing a flag value?

    The AI will follow your direct instructions. If you tell it to update a value, it will perform the action.

  • 04

    Can I use this to create new feature flags?

    No. The AI can manage values and segments, but it cannot create new feature flags from scratch.

  • 05

    How does the AI know which users are in a segment?

    The AI uses the segment tools to define and manage the groups, but the actual user attributes are managed within your ConfigCat setup.

  • More questions about ConfigCat? The Connector page answers them. See everything the ConfigCat Connector can do

Connect ConfigCat to Claude, Cursor, ChatGPT & more