AI in go-to-market isn’t a magic shortcut. It’s a precision engine, but only if your fundamentals are right. In this episode of GTM Hackers, Charles Brun sits down with Stephen Bates, Founder of Cabot Insights, to unpack what actually works when startups try to implement AI, automation, and GTM engineering.

They break down why most viral automation workflows fail in the real world. Teams move too fast, skip ICP definition, ignore data quality, and expect tools to fix broken foundations. Stephen shares how his team uses Clay and LLMs to enrich accounts, capture intent signals, and score ICP fit at scale, while keeping human judgment and feedback loops in the system so reps actually use the output.

Key discussion points:

- Why rushing AI adoption creates long term GTM problems

- Crawl before you run: data hygiene, CRM quality, and ICP clarity first

- How Clay enables micro segmentation, enrichment, and ICP scoring at scale

- Prompting strategy: specificity, strict outputs, and auditability for trust

- How to detect intent signals using real account signals

- Choosing models by ROI: lower cost scoring vs higher quality email generation

- Adoption and reinforcement: weekly AE feedback loops and enablement

Guest:Stephen Bates is the Founder of Cabot Insights, a GTM engineering agency helping high growth startups implement AI automation to improve outbound efficiency and lead quality. Stephen Bates brings experience from enterprise sales and the startup ecosystem, with a strong focus on ICP rigor and measurable ROI.

Follow GTM Hackers, share this episode with a GTM leader, and leave a rating if you want more deep dives on AI for go-to-market.