Work with me
Complex enterprise work, redesigned with AI
Most enterprise AI stops at a tool being deployed. The work underneath — the handoffs, the approvals, the exception that takes three days — carries on unchanged. I work with teams on the other part: finding the workflows where AI genuinely changes the economics, redesigning them end to end, and making the architecture around them safe enough to run.
Eighteen years inside SAP, finance, treasury and enterprise systems programmes sits behind it. That is where I learned what actually survives contact with a control environment, a cutover and an audit — and what quietly doesn't.
AI Workflow Opportunity Scan
Which of your workflows are actually worth redesigning with AI — and which aren't.
For a function or programme that knows AI should matter here, but not where to start.
We go through the real workflows in a function — not the process map, the way the work is actually done — and assess each one against the things that decide whether AI changes its economics: how often it runs, where the human hours go, how much of the input is unstructured, how repeatable the judgement is, what the exception rate looks like, what a wrong answer costs and who would have to own it.
What it covers
- A shortlist of candidate workflows, ranked, with the reasoning visible
- For each: automate, augment, agent — or leave alone, and why
- The data and system state each candidate would need to exist
- Control and risk implications, including who decides what
- An honest read on implementation complexity
You end up with: A written assessment you can take to a steering committee, and a defensible reason for the two or three you decide to pursue.
Ask about this →AI Workflow Redesign Sprint
Take one workflow apart and put it back together as it would be designed today.
For a workflow you already know is a problem: slow, manual, exception-ridden, or quietly expensive.
One workflow, end to end. We map how it runs now and where the time and the rework actually sit, then separate what must stay deterministic — calculations, limits, postings, segregation of duties — from what AI genuinely does well, and from the decisions that need a named human. Then we design the version that owns its own state, handles the exception path rather than ignoring it, and can be audited.
What it covers
- Current-state map with the time and rework where they really are
- The deterministic core that must not be handed to a model
- Where AI reads, drafts, classifies or proposes — and where it stops
- Decision rights: who approves what, and on what evidence
- The redesigned workflow, its exception path, and what it needs to hold
- A business case in cycle time, rework and risk, using your numbers
You end up with: A redesign your team can build against, with the trade-offs written down rather than discovered later.
Ask about this →Enterprise AI Architecture Review
A second pair of eyes on an AI use case or agent system before it meets your real landscape.
For teams with something working in a pilot and a nervous conversation ahead about production.
The hard part of enterprise AI is rarely the model. It is where the capability sits relative to the systems of record, what it may write and under whose identity, how its actions are logged and reconciled, what happens when an upstream system is unavailable, and how any of it is rolled back. This is the review that asks those questions while they are still cheap to answer.
What it covers
- Placement: read-only, propose-and-queue, write-through, orchestrating
- Integration with the systems you already run, and who owns which record
- Data access, identity and scope — what it can reach and why
- Controls, segregation of duties and the audit trail
- Human oversight boundaries: which decisions never become automatic
- Observability, failure behaviour and what rollback means here
You end up with: A written review with the risks named, the gaps prioritised, and the specific changes that would let it run in production.
Ask about this →How this works in practice
Scope, length and price depend entirely on the workflow and how much of your landscape is in scope, so I'd rather talk about the actual problem than publish a package price that fits nobody. Remote by default; on site where the work genuinely needs it.
I'll say plainly when something isn't worth doing with AI. A scan that comes back with "two of these, and honestly none of the rest" is a good outcome — the expensive version is the one where nobody says it.
What I can't offer you yet
No client logos, testimonials or "we saved €X million" numbers. This is a new practice and I'm not going to invent a track record for it. What I can offer instead is everything I've published: how I think about AI transformation, the workflow-design method I'd apply to your process, the architecture questions I'd ask, and years of field notes from the enterprise systems work behind it. Read a few and you'll know whether I'm the right person before we ever speak.
The free tools are the same thinking in a form you can use today without contacting anyone at all, and the products are where I build the ideas rather than only writing about them.
Get in touch
Tell me about the workflow: what it is, roughly how it runs today, and what makes it painful. That's enough for me to say whether I can help and which of the three above fits.