Stage 2 of 8 · 9 lessons · ~5 weeks
Prompting and model APIs
Treat prompts as code and model calls as an unreliable network dependency. Be specific, fix the output format and show examples, then learn one provider's API properly: system prompts, streaming, retries and a local model for cheap experiments.
Start: The chat messages APILessons· one lesson for each topic, in this order
Start with· follow one of these
- CoursePaidGrokking Prompt Engineering for Software Engineers
Design Gurus · Beginner · Subscription
Prompting patterns for engineers building features, not for chatting
- ToolFreeVercel AI SDK Documentation
Vercel · Intermediate
Free: the Foundations and Getting Started guides for prompts, generateText and streamText
Practise
Keep handy
Also worth your time
Few-shot, chain of thought, JSON enforcement and parsing failures in one place
Before moving on: Build a Next.js or Node.js endpoint with the AI SDK that turns messy text, such as a job post or an email thread, into a fixed output format, streams its reply to the browser, retries on timeouts, and runs against both a hosted model and a local one through Ollama.
Get a plain prompt with clear instructions, a fixed output format and two or three examples working before you reach for tools or frameworks; most weak outputs come from vague prompts, not weak models. Keep prompts in Git and rerun them on the same ten inputs every time you change a word.