Starter prompts
4 ways to start with AIEng.
RAG
→ Production-ready
▸ Preview prompt
Design a production RAG over our 50k-doc knowledge base — chunking, embeddings, retrieval evals, citations, fallback when retrieval is empty.
Agent
→ Tool use
▸ Preview prompt
Build a tool-using agent that triages support tickets — what tools, what guardrails, how I measure quality before shipping.
Cost
→ Half the spend
▸ Preview prompt
Our OpenAI bill is $40k/month. Lay out a plan to cut 50% with no quality drop — caching, routing, smaller models.
Eval
→ Catch regressions
▸ Preview prompt
Set up an eval pipeline that runs on every prompt or model change and blocks deploys on a drop.
What it does
Tasks AIEng ships every week.
AI features
- RAG with retrieval evals
- Tool-using agents
- Streaming + token caching
- Guardrails + safety
Operations
- Inference cost control
- Latency budgeting
- Eval + regression
- Model swap + A/B
Worked sample
A real AIEng chat.
Pairs well with