ChatGPT integration

ChatGPT is useful as an operator when it can call real tools instead of only drafting copy you paste elsewhere. Connected to Omentir, it can configure product context, create agents, inspect leads, and check send schedules while Omentir keeps LinkedIn credentials and safety limits in the workspace.

Finish Omentir setup: account, LinkedIn connection, My Product. In ChatGPT, add a custom connector with the hosted MCP URL. Complete the Omentir sign-in and Connect workspace approval. Enable tools in the chat when prompted, then ask ChatGPT to read context before creating anything.

• Connector URL: https://omentir.com/api/agent/v1/mcp
• Use workspace approval for chat apps
• Prefer list_agents before create_agent to avoid duplicates
• Keep reply tools limited to existing threads with draft approval

Typical sessions: refresh ICP language in My Product, spin up a classic lead finder for a new segment, start a Steal Customers agent against a short competitor list, or audit why discovery returned empty results by checking activity tools before inventing leads.

Do not treat ChatGPT as a place to store LinkedIn passwords or to bypass daily limits. Do not paste API tokens into chats you share. For scripts and local coding agents, create a revocable key on the API keys page instead of reusing a chat connector session.

Have ChatGPT explain the workspace back to you: product summary, connected account, existing agents. If that summary is wrong, fix My Product before you let it create outreach. A confident draft from a wrong profile wastes a week.

When you do create an agent, give ChatGPT the ICP or competitor URLs in the same message. Do not ask it to guess the market. Guessing is how you end up messaging the wrong titles at full pace.

ChatGPT is the right surface if you already live in that chat app and want to operate Omentir in language. If you want a coding agent to call REST from a repo, or you need a token you can revoke per script, create an API key and leave ChatGPT out of that path.