Services / AI Workflow Assessment
Assess · Start here
Find out whether AI is worth testing in one workflow
I talk with the people who do the work, walk through recent cases, and map where time is lost, errors happen, or judgment matters. In about one to three weeks, you get a workflow map and a one-page decision brief you can share with your team. It recommends whether to rehearse one AI use case, improve the workflow first, or park the idea.
Bring one workflow and the decision you’re facing. We’ll talk through where the work gets stuck and whether an Assessment would help. If there’s a fit, I’ll send a short proposal with scope, timing, and price within two business days.
What you get
Evidence you can share, and a clear recommendation
I bring the evidence and recommendation. You make the Greenlight decision: rehearse one AI use case, improve the workflow first, or park the idea.
01
Workflow map
How the work really happens: steps, tools, handoffs, judgment calls, and where time is lost or errors slip through.
02
One-page decision brief
Ranked pain points, a trust baseline, what is still unmeasured, and the limits any AI trial must respect.
03
A clear recommendation
Proceed, revise or park, with the evidence behind it, so your team can see why and decide.
From the sample decision brief
FICTIONAL TEACHING CASEBAYSIDE HOME SERVICES · CUSTOMER REPLIES
The workflow
Five dispatchers answer about 200 service requests a week. Each reply needs the customer’s history, technician availability and the right policy wording.
What the work showed
About 8 of every 10 minutes go to finding information. Writing the reply takes about 2. “Writing a reply takes a few minutes. It’s not the problem.” (Denise, dispatcher)
Ranked pain points
1. Finding the facts across three systems. 2. Getting policy wording right every time. 3. Urgent requests, such as a possible gas leak, must always reach a person.
Recommendation
Proceed: rehearse AI reply drafting with the history, schedule and policy shown beside each draft. A dispatcher reviews and sends every message. Rehearse, not deploy.
IN THE FIELD · CORPUS CHRISTI, TX
How the research works in practice
WHAT WE OBSERVED
During a disaster simulation for the City of Corpus Christi’s emergency management team, my team watched how people used our planning software alongside their existing tools, procedures and workarounds.
WHAT WE LEARNED
[WHAT THE TEAM LEARNED FROM THEIR TOOLS AND WORKAROUNDS]
WHAT CHANGED
[WHAT CHANGED IN THE INTERFACE AS A RESULT]
I bring the same research habit to every AI Workflow Assessment.
How it runs
The Assessment covers the first two turns of TURNS: Target the Mission and Understand the Humans.
STEP 1
Agree on one workflow
The workflow, your hypothesis, the people affected and the decision you need to make, captured on the Target Statement and Success Criteria Worksheet.
STEP 2
Talk with the people who do the work
Interviews across the sponsor and the team, a walkthrough of a recent case, and client-approved examples of routine work, exceptions and workarounds.
STEP 3
Map the work
Steps, tools, handoffs and judgment calls. Where rework happens, how staff catch errors, and what happens when one gets through. Pain points ranked, with a trust baseline.
STEP 4
Walk through the evidence
An annotated workflow map and a one-page decision brief with my recommendation (proceed, revise, park), reviewed with you and the workflow owner. You make the Greenlight decision.
Two scopes
Express or Standard
| SCOPE | Express | Standard |
|---|---|---|
| Price | $2,500–$3,500 | $4,500–$7,500 |
| Best for | One narrow question, or a sponsor who has already done much of the groundwork | A workflow with several roles, handoffs or exceptions |
| Example | One handoff with a known bottleneck, such as quotes that stall between sales and scheduling | Intake through dispatch, where several roles handle exceptions |
| Interviews | Two to three | Four to six |
| Observation | One recent-case walkthrough | Walkthrough plus review of routine work, exceptions and workarounds |
| Deliverables | One-page workflow map and decision brief (ranked pain points, trust baseline) with my recommendation: proceed, revise or park | Annotated workflow map and decision brief (ranked pain points, trust baseline) with my recommendation: proceed, revise or park |
| Readout | One session with the sponsor | One session with the sponsor and workflow owner |
| Timing | About 1–2 weeks | About 2–3 weeks |
| Your team’s time | [SPONSOR AND STAFF HOURS, EXPRESS] | [SPONSOR AND STAFF HOURS, STANDARD] |
| Minimum to start | A sponsor who can decide, two to three people who do the work, and one recent case. A redacted example or a walkthrough is enough. | A sponsor and the workflow owner, four to six people across the roles, and a few recent cases, including exceptions. Redacted examples are fine. |
Delivery is remote first. Extra locations, recruiting, travel, subscriptions and production development are scoped separately.
The decision brief
Three honest outcomes
Your starting idea is recorded as a hypothesis, not a conclusion. Each important finding is traced from observation to interpretation to requirement to the question that would validate it. The brief ends in one of three outcomes.
Proceed
Rehearse one AI use case.
The brief specifies the task, baseline, representative cases, human review and the evidence needed to authorize a limited pilot.
Revise
Improve the workflow first.
The brief names the handoff, ownership, inputs or process issue to fix, and who should own it.
Park
Stop or defer.
The brief explains what makes the opportunity premature, and what evidence or change would justify revisiting it.
A Proceed means rehearse, not deploy. Authorizing a limited live pilot is a separate On Air call (go, adjust, no-go) after the rehearsal. A well-supported Revise or Park is a good result: it saves you the cost of a pilot that was never going to work.
See it before you buy it
A sample decision brief
Bayside Home Services is a fictional 60-person company used in every TURNS course. The sample shows how the owner's first idea, AI for applicant screening, becomes a hypothesis to examine, what the dispatch team's work reveals, what is still unmeasured, and what would justify a rehearsal.
It is a teaching case, not a client.
Download the sample brief (PDF)BAYSIDE HOME SERVICES · DISPATCH
Terms on every proposal
Clear before kickoff
Access
You arrange staff access and examples before kickoff. Work pauses if confirmed interviewees and examples are not in place within 10 business days of kickoff.
Confidentiality and data
A mutual NDA is signed before any example is shared. I use only examples you approve. No client data goes into an AI tool without your written permission, and the proposal names any tool used.
Who decides
I bring the evidence and recommendation. You make the Greenlight decision (proceed, revise, park), and the selection stays provisional.
Questions leaders ask
We haven’t picked a workflow yet. Where do we start?
Start with the free Pilot Selection Scorecard. It helps you score your ideas and choose one. If you’re still deciding between two or three, bring them to the free call and we’ll narrow them down together.
When does a Solo Workflow Review make more sense?
When you run the workflow yourself or with a very small team. The Solo Workflow Review is a single 60–90 minute live session that ends in the same kind of one-page recommendation, without interviews across a team.
Can we start with redacted examples?
Yes. Redacted examples or a live walkthrough of a recent case are enough to start. A mutual NDA comes first, nothing goes into an AI tool without your written permission, and a live pilot is always a separate decision.
Do we need technical staff to take part?
No. The Assessment is built for business and product leaders and the people who do the work. Technical input helps but isn't required.
Will you recommend specific AI vendors?
Only when a tool clearly fits. Recommendations are vendor-neutral and based on fit alone. My public Tools Hub uses some affiliate links, always disclosed, and they never factor into client advice.
Is this the same as an AI Discovery Sprint?
Yes. Teams used to product discovery sometimes call it that. Either way, the deliverables are the workflow map and the decision brief described above.
What happens after a Proceed?
The brief is written so your team can run the rehearsal itself. If you’d like help, I scope the rehearsal separately, or I can help with a pilot planning package or a Human Centered AI Pilot. Teams that want to run the next one themselves often take the Five-Session course.