The AI context gap: why ChatGPT keeps giving you someone else's answers

A business owner I spoke to recently had done exactly what the LinkedIn gurus told him to. He asked ChatGPT to write a business case for expanding into a new region. What came back was polished, structured, confident and useless. It could have been written for any business in any region. He concluded AI was overhyped and went back to his spreadsheet.

He was half right. The output was useless. But not because the tool is weak; because it knew precisely nothing about his business. Not his margins, not his customer concentration, not the fact that his two best fitters won't relocate, not the regional competitor with a cousin on the council. Asked a generic question, these tools return the average of everything ever written on the topic. Impressive fluency, zero specificity. That is the context gap, and it is where most small business AI effort quietly dies.

Context is the whole game

Enterprises have spent the last few years, and considerable millions, learning this exact lesson. The firms getting real value from AI are not the ones with the cleverest prompts; they are the ones that got their context organised: what the business actually does, for whom, at what economics, in what voice, under what constraints. Feed that in and the same tool stops being a plausible stranger and starts being a well-briefed analyst.

The good news for a small business is that your context is small enough to capture. An enterprise needs a programme to do this. You need a disciplined afternoon or two: your numbers, your customers' own words, your pricing logic, what you will never do, how you sound when you write. Written down once, it works in every tool, every time.

Five habits that close the gap

First, never ask a generic question you would not ask a stranger. Give the tool your actual figures, your actual constraints, your actual customers. Second, make it argue against you: "give me the three strongest reasons this expansion fails" produces more value than any cheerleading. Third, use your customers' words as raw material; testimonials, complaints and enquiry emails carry more truth than any prompt you will write from memory. Fourth, verify anything that touches money, law or customers; the tools remain confidently wrong at times, and the cost of checking is minutes. Fifth, keep your context in one living document and improve it each time an answer misses, because the misses tell you what the tool did not know.

The test

Take the last AI output you were disappointed by. Ask yourself honestly: did it fail because the tool is weak, or because a smart human given the same brief would have produced the same generic mush? In my experience it is the brief, four times out of five. Close the context gap and the same tools that wasted your evening start earning their keep.