AI architecture · 90-day engagements
Most AI programs stall long before anyone admits which ones were never going to work.
I help CEOs and technology leaders separate the AI initiatives that will hold from the ones that will not, sequence the survivors, and build the operating discipline to run them without me.
The diagnosis
Is the AI portfolio real, or is it a list of pilots nobody will kill?
Most organizations do not have an AI problem. They have a list of AI initiatives, a handful of which matter, and no mechanism for saying so out loud. A 90-day engagement establishes which is which, sequences the work that survives, and puts the governance in place to keep the next list from accumulating.
See engagements- Portfolio triage
- Architecture and sequencing
- Operating discipline
Architecture before adoption
Adoption is an activity. Architecture creates an outcome.
Organizations rarely need another isolated AI campaign. They need coherent architecture, interoperability, governance, and the judgment to make the whole system perform under real operating conditions, including the conditions nobody modeled.
Read the full ideaFrom the field
From my LinkedIn writing
“The most valuable problems don’t announce themselves. They hide inside vague, throwaway statements that most people nod past.”
Reading the Vague Statement: Problem Solvers
I listen for the hedge: “fine,” “well enough,” “for now.” Then I find out what “fine” is costing. The complaint is usually a symptom. The work is to trace the symptom to the system underneath it, and solve the problem that actually matters.
Enablement
Decisions do not execute themselves.
A roadmap the team cannot run is a document. Harness Academy is the enablement layer of the work: a shared AI foundation with distinct paths for builders, solution architects, and leaders, built so the capability stays after the engagement ends.
Founder and Lead Architect
Evidence posture
Most of the work is private. The proof boundary is public.
Each case labels what its evidence can and cannot prove. Private implementations are separated from measured outcomes, deployed systems, and public artifacts so the claim never exceeds the evidence.
Read the evidence conventionImplemented capability
What the system was designed and built to do.
These are implemented capabilities, not claims of achieved business impact. The supporting case states the evidence boundary.
Governed research
implemented workflow capability
Coverage model
implemented operating capability
Continuous intelligence
implemented system capability
Career progression
A career of expanding responsibility.
The work moved from making systems perform to building the conditions that let organizations and leaders perform.
1998 onward
Hands-on engineer
Built the craft required to make complex systems work in the real world.
Expanding scope
Enterprise architect
Connected individual systems to business outcomes, operating realities, and long-term decisions.
Capability leadership
Practice builder
Established technical practices, reusable systems, and a national capability from the ground up.
Today
AI architecture and leadership
Builds the structures, governance, and people that let organizations make sound AI decisions closer to the work.