Blog
News, guides, and updates from Curtis Foster.
Multi-agent systems: when they help, and when one good agent wins
Multi-agent systems can divide complex work, but they also multiply latency, cost, and failure points. Here is how to decide when one good agent is enough.
What a solo founder learned shipping AI from 0 to 1
Building Ziggyloo and Zentoko alone changed how I think about scope, reliability, cost, and the work that turns an AI demo into a product.
AI memory architecture: the five layers behind reliable LLM products
AI memory is not one database or a longer context window. It is five separate layers, each with a different job, cost, and failure mode.
The eval loop is the product: measuring whether your AI is actually right
A production AI system is only as good as its measurement loop. Learn how to build AI evals that catch bad answers before users do.
Fix the corpus: a field guide to RAG data quality
RAG failures usually start before retrieval. Learn how to audit, clean, chunk, and test the corpus behind a reliable AI application.
How I build a self-certifying review loop for production AI
Production agentic AI needs more than a final answer. Here is how I build review loops that test evidence, correct failures, and know when to stop.
From technical program management to AI builder: what transferred, and what didn’t
Technical program management teaches you how to move complex work. AI building adds uncertainty, product judgment, and a tighter feedback loop.
The three numbers I watch on every AI product I ship
AI products fail in the gaps between quality, latency, and cost. Here is the practical dashboard I use before users find those gaps for me.
Hiring your first AI engineer as a non-technical founder: what to actually screen for
Your first AI engineer does more than write code. Here is how to screen for judgment, product sense, and the habits that keep an AI team sane.
Agentic pipelines: what a self-correcting review loop actually buys you
A self-correcting AI review loop does more than catch mistakes. It gives your agentic pipeline bounded recovery, better evidence, and clearer operating costs.
The distribution moat most AI startups ignore until it is too late
A strong AI product is not a distribution strategy. Here is how to build a repeatable path from first user to durable growth before paid acquisition becomes a crutch.
Why AI cost telemetry belongs in your MVP, not your Series A
Your MVP does not need perfect finance reporting. It does need to show what each AI action costs before usage turns into a surprise.
Subscribe to Curtis Foster
Get new posts in your inbox. No spam, unsubscribe anytime.