Dr. Wendy KimbrellEd.D.
Connect on LinkedIn
Every request enters through one front door, moves through four stages, and ends with a way of telling whether anything changed. This is the part most AI conversations skip.
01 / Assessment
A self-assessment tells you what someone believes about their own capability. It does not tell you what they can do. The distance between the two is the most useful number in the room.
A short self-assessment covers the skills the work requires, using ratings, rankings, a proficiency matrix, and a few targeted open questions. It is quick, and people are usually generous with themselves.
A conversational activity uses open responses instead of multiple choice, and no AI assistance is permitted. People show the skill instead of describing it. Someone who rated themselves near the top sometimes cannot answer the first question.
Currently showing: Rated at capability
The gap is not a failing. It is where the design starts.
Innovation is the stage almost no one names. It arrives when people trust the tools enough to use them beyond their intended scope, and it is the only honest finish line.
02 / The intake
Every request enters the same way, whether someone submits it themselves or on behalf of a colleague. There are ten steps, and they group into four distinct moves. Open one to see what sits inside it.
A business problem or an opportunity is framed before anyone reaches for a solution. Most requests arrive already shaped as a tool. This move takes them back a step.
Start with the work and the desired outcome, not with the technology.
Capture the problem, current process, systems, effort, frequency, priority, timing, and what success looks like.
Current state comes first, then the whole system, and only then a recommendation. Sometimes the answer is not AI at all. Sometimes it is a process change or a tool the organization already owns.
Understand the workflow, the people, the technology, the constraints, and the capabilities already present.
Look across the full system and identify root causes, dependencies, and downstream impacts.
Determine whether the answer is AI, automation, existing technology, process improvement, or some combination.
A practical first version is tested in the real workflow, not in a demonstration. If the person who raised the request does not feel ready to build it, I build it and then teach them what I did and why.
Build a practical first version with the people who own the work.
Validate usefulness, accuracy, workflow fit, and adoption.
The thinking is taught, not just the tool. Ownership stays with the team it was built for. They operate it, they improve it, and only then does anyone ask whether it should spread.
Teach both the solution and the thinking behind it.
Hand ongoing ownership to the business team so they can operate and improve it.
Evaluate for reuse, standardization, or broader deployment.
That is what moves an organization from use cases to capability.
03 / The evidence chain
Five layers, each answering a harder question than the one before it. Most reporting stops at the first, because the first is the easiest to count.
A single metric gives you a line. Capability requires the shape.
Where this continues
I write about this while it is still in progress. LinkedIn is where that happens.