Enterprise AI adoption
I build the system that turns AI access into daily work.
I work with leaders and operators to diagnose where AI belongs, rebuild workflows, govern what reaches production and measure the result.
- Former Microsoft Global AI Director
- Founder, Institute for Applied AI
- PhD in AI
The operator’s method
Diagnose. Build with the work. Scale what proves value.
-
01
Diagnose the work
Operating principle
Start with real usage, friction, skill gaps and the tools people are already building, rather than a platform choice.
The work
Interview operators and decision-makers with one fixed diagnostic. Map repeated needs, case complexity and where the current tools are being quietly worked around.
Output
An adoption diagnosis, use-case inventory and leadership proposal.
-
02
Build with the operators
Operating principle
The people who know the work design the workflow, the knowledge and the checking step.
The work
Create a champion network, peer-led learning, role-specific credentials and living tools people can use inside the workflow.
Output
Working workflows, a champion network and evidence of capability.
-
03
Scale what proves value
Operating principle
Use cases earn their way to production through clear gates. Weak ideas stop early; useful work is reused.
The work
Set hub-and-spoke ownership, governance, the knowledge feedback loop, role changes and a scorecard tied to business outcomes.
Output
A governed portfolio, operating cadence and value scorecard.
Selected result
Global adoption at Microsoft.
Copilot and AI adoption across a 2,000+ person support organization.
I designed a hub-and-spoke model connecting a central AI Innovation Hub with regional leads, managers, practitioners and a 65-person champion network.
More than ten use cases were assessed. Four moved to production.
Long-form conversation
Why AI adoption is an operating problem.
“70% of what makes AI work has nothing to do with the tools.”
-
01
Why imposed tools are quietly demoted
-
02
Why adoption follows real capability
-
03
Why value is measured by use case, not agent count
Ways to work
Four places an engagement can start.
-
01
Adoption diagnosis
A fixed diagnostic across leaders and operators, ending in a use-case inventory and a plan.
-
02
Operating model & governance
Decision rights, intake, review gates, regional ownership and the path to production.
-
03
Workflow & workforce redesign
Human-agent handoffs, manager expectations, role changes and a champion network.
-
04
Portfolio & value measurement
A ranked portfolio with baselines, evidence levels and a cadence for deciding what scales.
Team programs
Bring one real workflow.
Institute for Applied AI
Teams rebuild work they already do, test it against a written standard and record the result. No demo workflow. No generic prompt library.
-
01
Diagnosis
Choose the workflow, risks and baseline.
-
02
Build
Rework it with the team that owns it.
-
03
Capability Pack
Keep the workflow, test result and evidence.