Operational design for the AI era
How should your company run, now that AI exists?
Most AI initiatives automate the org chart instead of the actual work. I map how your operation really runs — then redesign it around what AI does well, and what your people must keep owning.
The problem
The workflow is the product. Not the software.
Companies are buying AI tools faster than they are redesigning the work those tools are supposed to change. So the pilot demos well, the rollout stalls, and nothing measurable moves.
Tools land on top of broken process
A copilot bolted onto a workflow with six handoffs and three approval queues saves minutes and leaves the bottleneck untouched.
Nobody mapped the real work first
The org chart is not how work happens. The workarounds, shadow spreadsheets, and knowledge in one veteran's head are where the hours actually go.
No one drew the line
Without an explicit boundary on what AI must not decide, teams either over-automate into risk or freeze and adopt nothing.
The engagement
One ladder. Every rung earns the next.
No open-ended consulting. Each stage is fixed in scope, priced on value, and ends with something an executive can act on.
Workflow Assessment
Weeks inside one operational area — interviews, observation, and workflow mapping. You get the opportunity map, the do-not-automate boundaries, and a prioritized roadmap.
Pilot Design
The first implementation, fully specified: platform selection, success metrics, governance rules, and the business case for what comes after it.
Orchestration
Implementation directed, measured, and iterated. Your engineers or a delivery partner build; I own the outcome, the measurement, and the next decision.
The line
Every operation has work AI should never touch.
Naming that boundary is not a caveat at the end of the project. It is one of the deliverables. It is also what makes the rest of the plan safe to approve.
The method
Map. Redesign. Measure.
Five phases, run in order. The first two are where a decade of field research does the work that AI cannot do for you.
Discovery
Interviews and direct observation. How the work actually moves.
Opportunity map
Eliminate, accelerate, protect — task by task, with hours attached.
Workflow redesign
The future-state process. Software comes after the process, not before.
Build & orchestrate
Pilot specified and directed. Platforms chosen to fit the workflow.
Measure
Hours saved, cycle time, first-response time, error rate, cost avoided. Reported in the language your board already uses — not model benchmarks or adoption dashboards.
Who this is for
Operations-heavy companies where the work is complicated.
The engagements that pay off share a shape: expensive skilled labor, heavy documentation, real regulatory exposure, and decades of institutional knowledge held by people rather than systems. Energy and utilities, industrial manufacturing, logistics, insurance, and the operational core of technology companies.
The person who signs is usually a COO, a VP of Operations, or whoever has been handed responsibility for "our AI strategy" without being given a method to produce one.
Start here
Where does the work actually go?
That is the question an assessment answers. If you have an operational area where the hours feel unaccounted for, tell me about it — a short conversation is usually enough to know whether this is worth doing.