[ WHAT I DO ]
I find the work that is quietly costing you, fix it, and prove the saving.
Most businesses do not have an AI problem. They have manual effort, broken handoffs, systems nobody fully adopted, and knowledge that lives in one person's head. AI is useful against all four, but only once you can see them clearly and measure them. That is the job I do, either inside a business as its AI or delivery lead, or as a fixed engagement when you want the diagnosis and the plan without a permanent hire.
THE FOUR PROBLEMS
The four problems I work on
Manual effort
People copying, chasing, checking, rekeying or reconciling information. Data entry, manual checks, spreadsheet controls, repeat admin. It rarely shows up as a line in the budget, which is why it survives.
Broken flow
Work stalls because ownership, approvals, handoffs or decision paths are unclear. Approvals that wait for one person, cases that bounce between teams, escalations nobody owns.
System gaps
Good technology exists, but the operation still runs on workarounds. Disconnected systems, poor adoption, duplicate records, reporting done by hand because the dashboard was never trusted.
Knowledge risk
Operations that depend on one person's memory or undocumented judgement. Tribal knowledge, single points of failure, training gaps, decisions that come out differently depending on who made them.
HOW I WORK
Find it. Fix it. Automate it. Prove it.
Find it means understanding how the work actually happens, using process evidence, operational data and what the systems really do, not the version on the org chart. Fix it means redesigning the process before anyone automates it; automating a broken process just makes it fail faster. Automate it means building the right thing for the job, which might be an AI agent, a workflow, an integration, a dashboard, or a small internal tool. Prove it means measuring time saved, cost removed, capacity created, risk reduced and speed improved, against a baseline taken before the change.
In practice that runs as four stages, and each one has to earn the next.
Diagnose
A focused sprint on one operation to quantify the opportunity and decide whether the change is worth making at all.
Design
The future-state workflow, the operating model around it, the controls, the data it needs, and an implementation plan sized against the team you actually have.
Build
Automation, AI agents, integrations, dashboards, internal tools or software, delivered in stages with a measurable test at the end of each.
Improve
Operational measures and review cycles after launch, so the change holds and keeps improving instead of drifting back.
WAYS TO WORK WITH ME
Ways to work with me
As a hire. An AI lead or delivery lead inside the business, owning the loop from finding the problem to proving the saving, and taking two or three improvements live with before-and-after figures. The six-week discovery described in Writing is the first part of that job; the rest is shipping and measuring.
Operational Performance Sprint. A fixed engagement on one problematic operation. You get a current-state diagnosis, a map of the opportunities, a future-state operating model, automation and AI recommendations, a prioritised roadmap, an estimated business case, prototypes where they help, and guidance for whoever implements it. It suits a process that is visibly slow, manual or expensive; work that several teams share with no clear owner; or complexity that has grown from system changes, acquisitions or new initiatives, where leadership wants a business case before committing. Duration and price depend on the shape of the sprint and are agreed before it starts.
Design and build. The follow-on from a sprint, or a defined piece of work in its own right: AI agent development, workflow automation, system integrations, low-code applications, operational dashboards and custom software.
Post-launch improvement. Setting up the measures and review cycles that keep a change working after the project team has moved on.
If you recognise one of the four problems above, the quickest way to find out whether I can help is a conversation about where the time is going.
GET IN TOUCHREAD HOW A SIX-WEEK DISCOVERY RAN