AI & Engineering 1 min read
What a production RAG pipeline actually costs to run
Retrieval sounds cheap until you price embeddings, storage and re-indexing at real volume. Here is the arithmetic we use before quoting a client.
Insights
Ideas, engineering notes and field reports
What we're learning building AI systems, MVPs and digital products — written by the team doing the work.
FeaturedAI & Engineering
AI & Engineering
Product & MVP 1 min read
Scoping an MVP when the client wants everything
The hardest part of an MVP is not building it. It is the conversation where you agree what is not in it.
AI & Engineering 1 min read
Why we build on Payload instead of a headless SaaS CMS
Self-hosted, Postgres-backed, and the content model lives in your repo. Here is the trade we made and when we would not make it.
Growth 1 min read
A 157 MB app on a 2 GB box: sizing servers honestly
We measured a cold Next.js build at 3.8 GB and the running app at 157 MB. That 24x gap decides your hosting bill.
Growth 1 min read
Core Web Vitals for content sites that actually rank
LCP is usually one image and one font. Fix those two and most sites pass without a rewrite.
AI & Engineering 1 min read
Field notes: shipping an AI agent that talks to real customers
Guardrails, evals and the escalation path matter more than the model. What we learned putting one into production.