I learned a lot doing the retrospective on my Census of Government LLM development sprint. I tried to share most of it in this article, what worked, what went wrong, and how I am changing my work with LLMs moving forward.
Getting good work from an AI agent on our Census of Governments API had more to do with planning, review and documentation than with how I worded prompts. The longer I kept steering a session past its opening instruction, the more often the agent admitted an error. 🔗
0 reposts0 likes