Insights / Start with the operating problem
Start with an operating problem, not an AI idea
A practical way to identify automation work worth doing before the conversation becomes about tools.
Most teams do not need another session on AI possibilities. They need a clearer way to decide which part of the work deserves attention first.
Describe the friction in plain language
Start with the moment that keeps recurring: a request arrives without enough information, an account owner spends an hour preparing for a routine call, or a status update has to be reconstructed from three different tools. If the problem cannot be described without naming a product, it is probably not ready to automate.
Follow the work to its actual outcome
Ask what happens after the frustrating step. Who needs the result? What decision does it support? What happens when the usual path does not apply? This reveals whether the opportunity is a small shortcut or a workflow that deserves to be designed end to end.
Choose a use case that can teach you something
The best first workflow is useful enough that people notice the difference, but narrow enough that the team can see its behavior and improve it. Give it a clear owner, a practical measure of progress, and an explicit place for human review.
A good automation idea earns its way into the roadmap by making a piece of work clearer, faster, or more dependable for the people who own it.