Solve the system, not the symptom.
I take senior technical and operating roles inside founder-led companies: the ERP, the integrations, the automation, the AI layer, and the discipline that holds it all together. I'm an operator, not a career consultant. The systems on this page run live businesses today, including my own.
Success-based fees/Two client seats/Waitlist when full
Your operations grew. Your systems didn't.
The ERP and the warehouse disagree by forty units and nobody knows which is right. Month-end drags into week two. Wholesale orders wait on someone remembering to check a portal. Every new channel added a tool, and every tool added another reconciliation.
The standard fixes make it worse: a hire doing manually forever what should have been automated once, or an agency that ships a deck and leaves before the invoices post wrong.
None of that is a people problem. It is a systems problem, and it compounds quietly until a big order, a big partner or a big month exposes it.
Senior technical and operating leadership, without the full-time seat.
Fractional CTO
Own the technical estate: the decisions, the vendors, the architecture, the roadmap.
- Platform and architecture calls: build vs buy, replatform vs repair
- ERP and integration ownership across NetSuite, Shopify, Amazon, 3PL and EDI
- Vendor and agency selection, with the incentives named
- AI systems designed with governance, not bolted on
Fractional COO
Make the operation predictable: measured, reconciled and run on exceptions.
- Order-to-cash, from intake through fulfilment, invoice, payment and close
- Fulfilment and 3PL operations, freight rules, inventory discipline
- Monthly close cadence and channel-level margin visibility
- KPI scorecards and an operating rhythm the team keeps to
AI operations layer
An agent layer over the stack you already have. The seams between systems are where businesses leak.
- Daily automated health checks across every integration
- Exception-based monitoring: stale orders, silent failures, drift
- Automated digests and triage that hand people exceptions, not busywork
- Approval gates, kill switches and audit trails as standard
How an engagement runs.
Audit the system as it actually is.
An evidence-backed review of the systems, data flows and failure modes, with observed defects kept strictly separate from assumption and opinion. You get findings you can hand to any implementer, including one who isn't me.
Rank the work by impact, effort and dependency.
A sequenced plan with decision gates, so you can stop at any stage and still have captured value. We agree the baseline and the measure of success here, before any build starts. Forecasts are stated as ranges with the assumptions visible.
Build modular, logged and tested.
Automation is replayed against historical data before it touches production. Guardrails go in first: approval gates, hard limits, kill switches, audit trails.
Operate on exceptions, not heroics.
Monitoring that treats silence as a problem, not a pass. Stale work surfaces itself. A weekly cadence keeps the system honest after I've stepped back, and the measures from step two prove whether the improvement held.
Built, shipped and running.
● Live
An ERP-centred operating layer for multi-channel commerce
- System
- A commerce operation selling through marketplace, DTC, retail and distribution channels, multi-entity, with a single ERP as the source of truth and fulfilment through integrated 3PL warehouses.
- Mechanics
- Order-to-cash automated end to end: custom scripts for billing and lot assignment, orders flowing to the 3PL without manual touch, freight rules applied automatically, and a day-by-day monthly close.
- Net effect
- Multiple channels and entities run daily by a small founder-led team, with routine work handled by the system and exceptions handed to people.
● Live · since May 2026
An agent layer that watches the seams
- System
- A portfolio of purpose-built AI agents over a live business stack: integration health, order-pipeline lifecycle, error triage and executive-priority monitoring.
- Mechanics
- Daily green/red status across 10+ integrations, with "no signal" treated as a failure rather than a pass. Every active order knows its pipeline stage and next expected signal; anything stale surfaces as an exception. Script-error triage cut from roughly 30 minutes of context-hunting to seconds.
- Net effect
- Months of unattended scheduled runs producing daily operational digests, governed by approval gates and audit trails.
2026 · UK
Audit to board-ready strategy for a UK contract-catering group
- System
- Engaged on a single underperforming page; expanded into a category-leadership review covering search, interface quality, competitive position and online reputation.
- Mechanics
- Live-site and competitor audit, evidence separated from assumption, then a 22-page two-quarter transformation proposal spanning brand and demand, AI-enabled kitchen operations, and fair site-performance management, with every forecast stated as a planning range.
- Net effect
- From first look to board-ready investment case in 48 hours, including the platform architecture and compliance requirements to execute it.
● Live
Risk-gated execution infrastructure
- System
- Live automated trading infrastructure: signal webhooks into a risk engine that gates every order before execution.
- Mechanics
- Hard loss limits, position caps, force-flat protection and a one-word kill switch. Nothing goes live without explicit approval; every action is journalled.
- Net effect
- The discipline your automation inherits: if a system can act with real money, it gets real guardrails.
The person behind it.
I co-founded a consumer products company in 2020 and still run its operations, supply chain, finance and technology day to day. The systems on this page are the ones running it. Before that I spent a decade in multi-location retail operations, including helping scale a repair chain from six locations to fifty.
Consulting gets a deliberately small slice of my week. I take it on because I like business itself: the mechanics, the numbers, the moment a fix holds. Two client seats keep that work serious without pulling focus from my own company.
How I take on work.
Success-based, agreed against a baseline.
Engagements are structured around results, not hours. We agree the baseline and the measure before work starts, and the fee follows the improvement. If the improvement can't be measured, I won't price it as if it could.
Only work I can improve and prove.
I take an engagement when I'm confident there's a measurable result to reach. That's what the diagnostic is for. If I can't see the path, I decline and say why, and you keep the findings either way.
Two seats, then a waitlist.
My own company comes first, and client work gets the part of my week I can protect. That caps it at two companies at a time. When both seats are taken, new engagements join the waitlist in order.
Where I'm useful, and where I'm not.
A good fit
- Founder-led company where operations have outgrown the systems underneath them
- Multi-channel commerce, ERP-centred stacks, or an operation drowning in manual reconciliation
- You want AI working in the business, with governance, not a demo
- You want the flaw in the plan named before it costs you
A poor fit
- You want a dev shop to staff a ticket queue
- You want guaranteed outcomes; I price against measured results, not promises
- You've already decided and want a consultant to agree
- The problem is a pitch deck, not an operation
I want rigorous thinking, not agreeable thinking, and I assume you do too. If an idea has a flaw I'll name it. If there's a better alternative it goes on the table before we commit.
Start with the system that hurts.
Email a short description of the operation and where it leaks: the reconciliation that never closes, the channel held together by one person, the integration nobody trusts. The first conversation is a diagnostic, not a pitch. If I'm not the right person for it, I'll say so. If both seats are taken, you'll get a straight answer on timing and a place in the queue if you want one.