Applied AI workflows: prompts, retrieval, and evaluation
A methodical walkthrough of a practical LLM workflow: prompt structure, light-weight retrieval augmentation, and a simple evaluation loop. The focus is on repeatable steps and how to prevent “looks right” outputs from slipping into real work.
- Build a prompt template and a “definition of done” checklist.
- Understand when retrieval is worth adding and when it is noise.
- Leave with a small evaluation rubric for consistent outputs.