Knowledge Systems: The New GTM Stack — Jeffrey Wang, Exa

Over a week off, Jeffrey Wang built an AI clone of himself. He analyzed 760 of his own emails to derive his voice, down to averaging 18 words and signing off with best rather than sincerely, then turned hundreds of past decisions into evals to calibrate the agent's judgment against his own. Anyone at Exa can now ask Jeffbot to draft a Slack message. Wang cofounded Exa, a search engine for agents, and his framing is that go to market is now an AI engineering problem. The product versus distribution argument he treats as settled: you need both. Underneath, go to market is a data problem, which means keeping a live model of your world agents can act on. Two interfaces and two agents make it concrete. An ICP dashboard classifies effectively every company in the addressable market, with anticipated spend attached, built by running Exa's embeddings over the web. Request Lens fires when something meaningful happens to a customer: a signup, a surge in searches, a sudden stop. The go to market team runs about a dozen agents in Slack against internal data. Wang closes on three principles: agent first requires API first, since agents cannot reach data without a programmatic interface; not everything should be a chatbot, because a consistent UI you learn once still beats a generated one; and buy versus build is a false dichotomy where what matters is whether the system is arbitrarily customizable. In the Q&A: Jeffbot reads and writes only when Wang calls it, and can only draft for anyone else. Speaker info: - https://x.com/jeffzwang - https://www.linkedin.com/in/wangzjeff/ Timestamps: 0:00 - Exa, and why engineers should care about go to market 2:56 - Go to market is a data problem 3:58 - A live model of your world that agents can act on 5:18 - The ICP dashboard: classifying your whole market 6:59 - Request Lens: alerts when something matters 7:41 - A dozen agents in Slack, and the go to market spend 8:22 - Jeffbot: cloning yourself from 760 emails 10:19 - Agent first means API first 12:01 - Buy or build is a false dichotomy 13:36 - Q&A: security boundaries and the FDE model