Media/Maven lessons
Today, watch us build and deploy a GTM agent
Building an AI GTM Engine/Michael Slawson & Benyamin Holley/
Live build of an inbound reply agent. They scope backwards from the outcome, then use Exa, LeadMagic, Lemlist, OpenRouter (Sonnet), webhooks, Vercel, and UTM parameters. Scaffold in Cursor, test the webhook, wire a Slack bot so the team can talk to it without logging into a CRM. The stated world: almost nobody should have to live in the CRM. Human-in-the-loop stays for warmth and qualification.
Notes
- 01
Work backwards from the outcome, then pick tools. This one is an inbound reply agent, not a generic chatbot.
- 02
Stack in the session: Exa, LeadMagic, Lemlist, OpenRouter/Sonnet, Vercel webhooks, UTMs.
- 03
Host it so you can push changes. Multiplayer means the people who use the bot hear about the change.
- 04
Slack is the interface. The point is fewer CRM logins.
- 05
Reliability still wants a human on warmth and qualification. They review output and prompt feedback live.
