The Workshop Invitation¶
After two weeks of posting about lib-pcb — the parsing complexity, the weekend experiments, the velocity numbers, the testing discipline — I posted something different.
Not more proof. An invitation.
The post¶
I'm planning a small, focused workshop in late February for developers interested in building production systems with Claude Code.
Not a talk. Not a demo. Working sessions where we tackle real verification problems, build actual testing frameworks, and figure out how to trust AI output in production.
The focus: - Building guardrails that actually work - Verification patterns for different domains - Testing strategies that scale - Working on YOUR code, not toy examples
This would be founding cohort stuff — you'd help shape what this becomes.
If you're working with Claude Code (or want to), and you have a specific pain point around trusting AI output, I'd love to hear: - Your tech stack (1-2 words) - Your verification challenge (1 sentence)
That was it. No pricing. No registration link. A question and an invitation.
What happened¶
25 likes. 19 comments. Both the best numbers any LinkedIn post had produced.
But the numbers aren't the story. The comments were: developers and engineering leaders showing up with specific tech stacks and specific pain points. Rust + egui. Java. TypeScript + GIS applications. Plugin marketplace development. Each one a real problem, not a generic expression of interest.
Jon Petter Hjulstad was starting a new company in renewable energy. Java stack, four-person team, needed to build a production system efficiently from day one. He sent a DM five days later. We met on February 3. He signed a contract on February 12.
Ola Prøis was building Ferrite, a markdown editor with 900+ GitHub stars. Already running 100% AI-coded production software. His specific question: better workflows for trusting AI output in production without slowing down.
Kent Vilhelmsen, CTO at ABEL, replied in Norwegian: "men JA dette vil jeg gjerne være med på!" — yes, enthusiastically, count me in.
Daniel Bentes was working on a plugin marketplace and described a cross-project testing standardization problem. He didn't end up in the workshop — he became the first Synthesis pilot user instead.
Six qualified leads from one post. 43% of commenters converted to pipeline. One contract signed within three weeks.
Why this worked and the technical posts didn't¶
The technical posts built credibility. They proved the approach was real. But credibility alone doesn't produce conversations — it produces silent appreciation.
The workshop invitation changed the frame from "here's what I built" to "here's a problem you have, and here's how we'd work on it together." The specific CTA (tech stack in two words, challenge in one sentence) wasn't just friction reduction — it filtered for seriousness. Someone who takes the time to articulate their tech stack and their specific pain point is someone with a real problem.
"YOUR code, not toy examples" was the phrase that came up repeatedly in comments as the reason people were interested. Generic courses work with generic examples. A workshop that addresses your specific codebase and your specific verification challenge is a different proposition.
The other factor: "founding cohort" positioning. Not open enrollment. A specific small group helping shape what this becomes. That framing attracts people who want to contribute, not just consume.
The sequence in retrospect¶
Looking back at the two weeks: Jan 15 (MPN parsing complexity), Jan 19 (weekend experiment), Jan 21 (velocity proof), Jan 23 (testing discipline), Jan 25 (invitation).
None of those earlier posts drove direct leads. But they created the context that made the invitation land. By the time someone read the workshop post, they'd already seen evidence that the methodology produced real results under real conditions. The invitation wasn't a cold pitch — it was a natural next step for people who'd been following the work.
That's the pattern: technical credibility first, then invitation. The credibility phase can't be skipped. But it also can't be the whole strategy.
Originally posted on LinkedIn, January 25, 2026. The post that started the workshop pipeline.
This post is part of the AI-Augmented Development series.