Automation & AI

Give the machine your context, not your hopes

A general tool gives generic output. The same tool, given your material, gives your output. The difference is not the model.

The most common disappointment with AI in a small business follows a predictable shape. Someone tries it, the output is plausible and generic, and the conclusion is that the technology is overrated. The tool was never the variable. What it was given was.

What context actually means

Not a clever instruction. Material: how you qualify an enquiry, how you price, who your suppliers are and what they charge, what you will and will not take on, the tone you write in, the objections you hear and how you answer them.

Most of that already exists in your business, scattered across people's heads, old emails and half-finished documents. The work is not creating it. It is collecting it once into a place a tool can read every time.

Documentation stops being defensive

Operating documentation is usually framed as insurance — protection for when someone leaves or is unavailable. That framing is why it never gets done: it competes with revenue work and always loses.

Given context to load, the same documents change character. They become the thing that makes every drafted reply sound like you rather than like anyone. One investment, two returns: the business explained once instead of re-explained daily.

How to tell it is working

The test is not whether the output is impressive. It is whether someone who knows your business reads it and cannot tell it was drafted. If they can, the context is thin, not the model.