The blog
Notes from production.
LLM agents, RAG, and multi-agent pipelines — written from systems with real users, not demos. Everything here is running somewhere on this site.
- 10 min read
How to Build an MVP That Ships in Weeks, Not Months
A working method for scoping and building a minimum viable product that reaches real users fast — what to cut, what drives the cost, and the failure modes that turn a six-week MVP into a six-month one.
Read it → - 5 min read
How to Build an AI Agent with Claude (Lessons from a Production Legal Pipeline)
A practical guide to building an AI agent with Claude — from a minimal tool-using loop to the multi-agent architecture running FASL, a production legal AI drafting Moroccan appeal memos.
Read it → - 5 min read
LLM Agent Architecture: What Actually Ships in Production
The anatomy of LLM-powered autonomous agents — planning, tools, memory, and orchestration — and how single-agent, pipeline, and supervisor architectures compare, drawn from systems running in production.
Read it →
Building with LLM agents?
I take on a small number of projects — AI-heavy, Arabic-first, or zero-to-launch products. Tell me about yours.