Getting Better Results (Steerability)

Lesson 1What steerability means

Prediction and memory explain what is happening under the hood. This module is about the part you actually control: your instructions. Steerability simply means how much your input shapes the output — and it varies a lot depending on how you ask. “Make this better” gives the AI almost nothing to work with; “shorten this to 100 words and make the tone more formal for a client email” gives it a clear target.

Lesson 2Be specific about the outcome, not just the topic

Instead of “write about our new product,” try “write a 3-sentence Instagram caption announcing our new product, exciting tone, aimed at young professionals in Accra.”

Real-life example: Kojo asks, “write about our new phone cases” and gets a generic paragraph. He rewrites it as “Write a 2-sentence WhatsApp status caption announcing new rugged phone cases just arrived, playful tone, targeting students in Winneba, mention the price is GHS 45.” The second version comes back sharp and ready to post.

Lesson 3Give examples, break tasks into steps, say what not to do

• •Give examples when you can. Showing one example of the style you want steers output far more reliably than describing the style in the abstract. Abena pastes a well-performing past customer email and says “write my next follow-up in this same tone and style.”
• •Break big tasks into steps. Kofi asks for the executive summary of his tailoring business plan first, reviews it, then asks for the market section, then the budget — steering the whole way instead of receiving one unfocused document.
• •Tell it what NOT to do. Esi adds “avoid technical jargon, explain it like you're talking to someone who has never used a smartphone app” to her mobile money fraud explainer, and the result becomes dramatically clearer.