Prompt engineering
Designing the input is designing the behaviour: instructions, context, examples and output contracts that make LLMs reliable.
In one line
Prompt engineering is the craft of specifying an LLM's task so precisely — role, instructions, context, examples, output format — that the model's behaviour becomes predictable enough to build on.
How it works
The reliable toolkit is small: a system prompt that fixes role and rules; few-shot examples that show the pattern rather than describe it; structured output contracts (JSON schemas) so downstream code never parses prose; explicit instruction to say "I don't know" and to ground claims in provided context; and chain-of-thought room for tasks that need reasoning. Treat prompts as code: versioned, tested against an eval set, changed deliberately — vibes-based prompting dies at the first model upgrade.
Where it shows up in digital health
Every clinical AI surface is mostly prompt + grounding: discharge-summary drafters, coding suggestions, Vaidya's persona ("teach digital health, cite Kosha, adapt depth to fidelity, never give patient-specific medical advice"). In health the prompt is also a safety artefact — scope limits and refusal rules live there, which is why it gets reviewed like one.