AI in practice
Give AI a job. Then check the work.
Start with a useful outcome, give it context, and keep judgement in the loop.
I am interested in what people can build with AI. The useful starting point is a business task: making an offer clearer, organising customer questions, drafting a first version or understanding where a process breaks.
A tool cannot supply the context you have not given it. Explain the customer, the objective, the facts it can rely on and the constraints it must respect. Ask it to name missing information instead of quietly filling the gaps.
Then decide what a good result looks like. If the job is to draft a follow-up, a longer message is not necessarily better. It needs to answer the customer, preserve the facts and make the next step clear.
This is how I connect AI to CLIMB. Understand the problem. Connect the right context. Make a practical plan. Produce the work. Check what happened and improve it.
Start with one repeated task. Use anonymised examples, review the output and compare it with your existing approach. Keep the part that helps. The aim is a better way to work, with judgement still attached.