Last week, I talked about how to write a good “prompt” by not writing it yourself. Instead, talk over the needs with your AI and let your AI draft it.

Honestly, prompt engineering is becoming an outdated concept, but the idea behind it is more valuable than ever. Today people call it “context engineering” — which is just a fancy way of saying: give your AI the right background information before asking it to do the work.

In the AI programming world, developers write careful planning documents with context, input/output formats, constraints, and rules. This guides their AI to stay on track, complete the project, and limit hallucination.

If you ever heard people mentioning “CLAUDE.md” or “AGENTS.md” — they are simply text files that document this planning. That’s it. Nothing magical. As sensory and consumer scientists, we don’t need anything as complex as what developers write. BUT! Even a simple planning document can significantly increase our success rate when working with AI.

So what should be in the planning? No need to remember a formula. Just ask your AI! Or better yet — just talk to your AI about your project, and then ask it to organize the conversation into a planning doc.

Here’s what mine would looks like for a consumer study:

• WHO I AM — sensory scientist at a beverage company, using R and Excel
• WHAT THIS PROJECT IS ABOUT — CLT study, n=150, 4 prototypes vs 2 competitors
• WHAT I NEED HELP WITH — ballot design, palate cleanser protocol, serving order, data analysis, writing a summary for the R&D team
• HOW I WANT ANSWERS — plain language, R-specific, flag anything that might bias results
• WHAT NOT TO DO — don’t make assumptions, ask if you’re not certain, don’t over-complicate the stats

Yours will look completely different — and that’s the point. It should be about YOUR project, YOUR tools, YOUR constraints.

Try it on your next project. You’ll be surprised how much better your AI performs when it knows what it’s working with.