Everyone wants to run GenAI workflows. The problem is they're running them on commoditized data, team sizes, job titles, basic firmographics  and wondering why nothing converts. Joe Lehr, Director of GTM Engineering & Innovation at Primary Ventures, works across a portfolio of companies from pre-seed through Series D, backed by a $625M fund five, and he sees this mistake at nearly every stage.

Joe gets specific about how he builds out GTM infrastructure from scratch: which signals actually move deals, how Parallel.ai powers Clay's CLAGIN for enrichment at scale, and what it actually takes to move a sales team from single-player AI experiments to a shared multiplayer system built on Supabase and MCP. He also pushes back on a few widely held beliefs, including whether deep hyper-personalization still works, why your first GTM hire should not be a 23-year-old, and why CS may be the most under-automated function in the stack.

Topics Discussed:

  • Non-commoditized signal mapping before any AI motion is built

  • Why creativity, not tooling, is the actual constraint at early-stage companies

  • The right first GTM hire: what to look for and what to avoid

  • How Parallel.ai powers Clay's CLAGIN for signal enrichment at scale

  • Mid-funnel data capture and why multi-opportunity edge cases break most tools

  • Build vs. buy: taking vendor sales calls before deciding what to build

  • Going multiplayer with Supabase, MCP, and structured account objects

  • Why deep hyper-personalization is losing ground to smarter segmentation

  • CS as the next frontier for agentic automation, expansion, and churn scoring

  • Observability gaps when vibe-coded internal tools skip engineering involvement

Rss
Podden och tillhörande omslagsbild på den här sidan tillhör GTM Council and Frontlines.io. Innehållet i podden är skapat av GTM Council and Frontlines.io och inte av, eller tillsammans med, Poddtoppen.