Most technical buyers evaluating an AI coding tool have already been oversold once by a wrapper dressed up as a platform.

The Pitch: Where I'd Start

I'd start with exactly what I already drafted: a first technical case study that leads with the governance objection before the benchmark, built from what's already public on cosine.sh instead of a generic vision statement. Read the draft →

More broadly, I'd write technical narrative that treats skepticism as the starting point, leading with the benchmark and the failure modes before the vision statement. I ship AI-assisted systems daily myself, so I can speak to agentic workflows from having actually built them.

I'd translate dense technical territory into content that respects the reader's intelligence. And I'd produce case studies that lead with the number a technical buyer actually checks: a benchmark score, or a specific failure mode fixed.

The Proof

I drafted Cosine's first technical case study, built entirely from what's already public on cosine.sh, plus the reasoning behind why I structured it that way. Read the draft →

I also built the Marketing Agent Skills library, 81 skills in daily use, 34 open source on GitHub. I write the MIA Newsletter every week for over 5,000 marketers and founders, at the same level of operating detail. And I built the Prospect & Outbound Engine, a self-built cold outbound pipeline running at real scale.

Let's talk about Cosine.

If this is the right problem for you, I'd like to talk about it directly.