Your go-to-market team should be building AI agents, not just using them

Jeff Wang argues that go-to-market is fundamentally a data problem, not just a sales or marketing function. He presents Exa's internal stack: an ICP dashboard that classifies every company in their TAM, a Request Lens for real-time customer signals, and a suite of AI agents used by their

GTM team. The most noteworthy claim is that GTM teams can now treat their workflow as an AI engineering problem. Exa's forward-deployed engineers both run deals and build the agent tooling that supports them, collapsing what used to be two separate roles into one. Wang also built a

personal AI clone trained on 760 emails and past Slack decisions. However, the talk leans heavily on Exa's own product being the search engine that powers these systems, creating an obvious marketing advantage. The claim that 'anyone can do this' glosses over the significant engineering investment required

to build custom dashboards and train decision-making models. The tradeoff is clear: deep customization versus SaaS procurement. Wang recommends maximizing customizability through APIs and MCP, but this assumes an engineering capacity most organizations lack. The organizational implications are real, but the evidence is mostly anecdotal and self-referential.