What an experienced builder would actually do — applied, post by post, to the problems people are bringing to AI agents right now.
Real founder and operator questions pulled from Indie Hackers, Reddit, X, LinkedIn, and podcasts.
Drawn from my own day-to-day AI exchanges and decisions — augmented with outside research and reviewed before publishing.
Problem named. Common flaws. The approach an experienced builder would defend. Provenance line at the foot.
From solo builders crossing the prototype wall to experienced teams answering for AI-shipped code. Each post tags to one (or more) of the six situations below.
A composite operator's stack laid bare — what AI-native tooling could replace this year, what needs to wait, and the order to do them in without breaking the business.
A composite founder scenario: a working signup flow shipped in a day that quietly stops scaling. What ships, what fails, and what a production-grade equivalent looks like.
A composite change-order, drawn from patterns across real dev-shop relationships. The agency approach, the in-house approach using AI-native tooling, and the conversation that has to happen first.
Same scope. Same week. Two paths. How that decision tends to play out — and where the contractor approach is still the right call.
What to do in the first week before anyone starts talking about modernization. The cheap, honest steps that change the shape of the conversation.
Each post doesn't start with me writing about a problem from scratch. It starts with a structured read of how I actually think — across agent transcripts, plans, reviews, commits, and decisions from my day-to-day founder work: the same working corpus Remaining Viable studies. That extracted approach is then applied to a publicly-sourced problem, augmented with external research where the problem warrants it, and reviewed before publishing.
This matters for two reasons. First, the approach in each post won't be self-asserted — it will be grounded in observable patterns from work that's already shipping. Second, the same analytic pass will produce signal for Data, particularly the Cognitive, Epistemic, and Judgment dimensions. One instrument, two directions.
Each post will carry a provenance line at the foot: the corpus window analyzed, external sources consulted, and the review pass that preceded publication.
Want more than the posts offer? Get the Playbook or become a mentee.
New Resources, Research, Data readings, and Playbook editions — this list is the only announcement I send.