Work I take on

The question nobody asks first

Who carries the risk when the tool is wrong?

The standard consulting arrangement is: the consultant delivers a slide deck or a demo, the client absorbs the real-world risk. When the AI summarizes a contract wrong, when the automation routes the wrong packages, when the dashboard surfaces a metric that nobody in the field trusts, the consultant is already on the next engagement. The client is the one holding the bag.

I do not work that way. I carry mail for a living. I have run quality control on a factory floor, coordinated clinical research, and built tools that live and die on whether an operator trusts them at 6am with cold hands. If a thing I build does not survive first contact with the person doing the work, I am the first one who feels it.

That is the posture I bring to paid work. Liability framed up front. Scope narrowed to what can actually be delivered. No handoff to an implementation partner. The person scoping the work is the person doing the work.

Practice area one

AI integration, honestly scoped

Most "AI integration" engagements are demos of what the model can do, not audits of whether the model should do it. The questions I answer first: what does the workflow actually look like in the field, which steps are safe to automate, which steps require a human reviewer to carry the liability, and what does failure mode look like when the model is wrong and nobody caught it.

Typical engagement: a single workflow, scoped to one or two real users, instrumented to catch the failures the vendor demo will not show. Output is a working prototype plus an honest writeup of where the model is trustworthy, where it is not, and what it costs to keep it trustworthy.

I do not sell "AI transformation." I sell one workflow at a time, with the parts that do not belong in a model carved out and left in human hands.

Practice area two

Legacy knowledge capture

Every organization with operators has the same problem. The person who knows the route, the machine, the account, the workflow, is about to retire, or already did. Their knowledge is not written down anywhere the institution can find. Training budget disappears into the gap.

I build the tools that pull that knowledge out before it leaves. Wiki systems with revision history so no contribution is ever lost. Interview and capture workflows designed for people who do not think of themselves as knowledge workers. Search and retrieval built around how the knowledge is actually used, not how a CMS wants it organized.

The public example is routelog.wiki: a wiki for USPS mail carriers, built by a carrier, that has 21 real users documenting stops, hazards, and route quirks that would otherwise walk out the door at every retirement. Same pattern applies to any organization where institutional knowledge outlives the institutional memory.

Practice area three

Operator-facing tooling

Software built for the person using it tends to get used. Software built for the person buying it tends to get abandoned. The gap between those two users is where most B2B tooling budgets go to die.

I build tools that close that gap. Mobile-first because the field is not at a desk. Offline-tolerant because the field loses signal. Stripped of the features that only exist to make the purchasing deck look good. Instrumented so the person doing the work can see that the tool is honest.

This is the bridge between "we bought software" and "people use software." Small engagements, tight scopes, written back up as case studies so the next client can see how the last one went.

How engagements run

One scope at a time

First conversation is free and by email. If the scope is something I can do, I propose a fixed fee for a narrow first deliverable plus a writeup of the failure modes. If it is not something I can do, I will say so and point you at someone who can.

I do not sell retainers. I do not sell "transformation." I do not sell anything that cannot be delivered and written up inside a single engagement. If the second engagement does not make sense, there is no second engagement.

Every deliverable ships with a writeup published on this site: scope, logic, fragility, learning. That writeup is the artifact the next client reads to decide whether to hire me.

Get in touch

If the work is in the field

Looking for a solutions architect or AI integration consultant who can translate between the people doing the work and the people buying the software. Open to conversations in GovTech, logistics, local MSPs, clinical operations, and B2B enterprise architecture.

Your email is not stored, tracked, or shared. This site does not use cookies for advertising. Minimal analytics are collected (page views with hashed, non-reversible IPs) for site improvement only.