Identify AI-enabled solutions.
Only after the problem is clear: practical ways AI could help this specific business.
This step is deliberately third, not first. Solutions matched to a diagnosed problem are useful; solutions chosen before the problem is understood are shopping. So the conversation turns to tools only once the constraint is named and tested.
The question is never “what is the best AI tool?” It is “what would actually fix this constraint, in this business, with this team, at this budget?” Anything that fails that test is set aside — even fashionable ideas.
What counts as a good solution here
- It attacks the diagnosed constraint directly, not a side problem
- It fits your systems and your people — no IT team required, no setup that eats the savings
- Its cost fits your budget, and its return can be measured in weeks, not vibes
- It has an owner inside your business and a weekly time budget that is real
What gets set aside
Generic tool lists. Ideas that need staff you do not have. Automation that removes the human touch exactly where your customers want one. A solution that nobody on your team can run is not a solution — it is a subscription waiting to lapse.
How it shows up in a real consultation
In the simulated physical therapy consultation, this step produced five opportunities aimed at the diagnosed constraints: automated waitlist and cancellation filling, a measured restart of an AI scribe pilot, referral intelligence, eligibility and authorization pre-checks, and a discharge-to-reactivation engine. Each one maps to a specific leak found in step 2. (Simulated example, not a real client.)