Guide · 7 posts

Ship LLM features that hold up in production

Calling a model in a request handler works in a demo. In production you need queues, retries, retrieval you can trust and code you can upgrade. These posts cover each piece.

Step 1

Run AI work reliably

Move model calls out of the request.

Step 2

Answer from your own data

Retrieval that cites its sources.

Step 3

Build and maintain

Starters and upgrades.

Step 4

Work with coding agents

Make AI coding tools follow your rules.

Want an AI feature built and run properly?

I build this for clients. Send the problem and a rough timeline.