Category
AI Product Building
Practical posts on shipping AI products: architecture patterns, evaluation frameworks, pricing models, multi-model orchestration and UX design.
Category
Practical posts on shipping AI products: architecture patterns, evaluation frameworks, pricing models, multi-model orchestration and UX design.
Showing 1–12 of 28 articles in AI Product Building

Six production chat surfaces, a habit of breaking every AI chat in the wild, and the defence-in-depth stack that keeps your prompts contained.

Product teams reflexively strip onboarding friction. Intentional friction that helps users understand why the product is for them increases conversion.

Enterprise software encodes decades of domain knowledge across every architectural layer. Vibe coding can't shortcut what took thousands of people 25 years to accumulate.

Chat is the wrong interface for AI agents in professional software work. A well-written issue is a better agent instruction than any prompt.

Everyone is asking which AI agent is best. The real question is which platform agents will work from. The answer is whoever owns the queue.

Made redundant with a newborn. Built a production SaaS in 10 weeks anyway. Vibe coding works, but not the way anyone's selling it.

Engineering teams spend more on AI tokens than junior salaries. The cost structure of building software has inverted and most finance teams missed it.

Zillow, Redfin, and Homes.com have all bolted conversational UI onto their portals. It's an impressive technical feat that completely misunderstands how people want to buy houses.

When AI writes the code, green CI isn't enough. The new discipline is understanding and defending the choices the model made — not just the ones you made.

The 'never rewrite' doctrine was based on rewrite cost. AI has collapsed that cost to days. Pre-launch rewrites are now a product strategy, not a failure.

Per-seat pricing is dying but the replacement is not simple. A practical framework for AI pricing that covers usage-based, outcome-based, and hybrid models.

Going from zero to end in hours sounds like progress. It's also how you ship a product nobody can navigate. The real skill is knowing when to stop.