Garden.
Notes, prompts, and playbooks — grown in public. Some seedlings, some evergreens, all things we've learned building AI systems for real teams.
Arabic-first AI, not Arabic-translated AI
Why models trained-then-translated fail Arab institutions — and what to demand instead.
AI literacy is a right, not a privilege
The case for teaching AI to people who will never become clients.
Give the team back the week
The only AI ROI metric worth defending in front of a minister.
Small models, sharp jobs
When a 7B model beats a frontier model, and how to know.
Context is the product
Prompts are cheap. The context you feed the model is the moat.
Taste is the bottleneck, not generation
Any team can generate a hundred options. Selecting the right one is the job.
Eval-driven design
Treat prompts like code. Version them, test them, delete the ones that fail.
Human-first defaults for AI systems
Five defaults we set on every deployment before anyone touches a prompt.
Procurement for AI, in plain language
A short RFP checklist for institutions buying AI without getting sold to.
Sovereignty is a workflow, not a slogan
What data sovereignty actually looks like once the press release is filed.
Train the trainer, not the tool
Why capacity building beats deployment every single time.
Voice guardrails for AI content
How to encode taste as a rubric so agents do not drift into slop.
The zag rule
If your competitor could run your prompt and get the same output, you don't have a brand.
Human in the loop, not on the leash
Where humans belong in an AI workflow — and where they get in the way.
Naming things with LLMs
A tiny workflow for generating names that don't feel generated.
The taste engine
Building an internal ranking model for creative output.