What Should Your Company Do Next With AI?

Most companies asking this are already doing something right. Someone drafts proposals with ChatGPT. A team built a shared assistant. Marketing keeps a prompt library.

So the useful question is not how advanced you are. It is which of three approaches fits the work in front of you. Here is how I tell them apart, using the same example for each.

Different workflows in the same company can need different approaches. A single person can benefit from customization, and a ready-made tool can serve the whole company.

How to choose

• Start with the tools people already have. If an existing tool handles the task well enough with reasonable instructions and review, that is your answer.
• Customize when results depend on your templates, knowledge, or standards. Giving the system that context once buys more consistent results with less repeated setup.
• Consider engineering when the work needs connected systems, clear controls, and someone to keep it running, and the improvement you expect justifies the cost of building and maintaining it.

Three signals tell me a piece of work is worth a closer look:

• People keep supplying the same instructions, or correcting the same mistakes.
• Meaningful time goes into moving information between systems by hand.
• It matters enough that someone would own it and measure whether it improved.

None of those is a reason to build on its own. They are reasons to look harder at the next approach. Often the honest answer is better context, or a change to the process itself, rather than software.

Talk it through

Bring a task, a recurring problem, or a question about whether your team is getting anything out of the AI it already has. We will work out which of the three fits, including when the answer is that you already have what you need, or that the next worthwhile investment is in your people rather than in software.