Claims intake, credit control, order entry, contract renewals. I build always-on agents that do the work end to end, keep a person on every decision that matters, and report what they save. For mid-market companies of 50 to 500 people, from first process to production in weeks.
Serial founder, four exits. PhD. Oxford Executive Leadership.
The money is being spent. The pilots are running. The board slide has a roadmap. And the P&L hasn't moved.
of CEOs say AI has delivered no significant benefits
Only one in eight report improvements to both cost and revenue. The problem isn't the technology. It's how it's being deployed.
PwC Global CEO Survey, 2026
from working demo to production-grade system
The gap between an AI demo and a system that runs unsupervised at scale is enormous. Bridging it takes someone who has done it before, not someone who's read about it.
Observed across dozens of AI engagements
of UK firms using AI use it extensively
35% of UK firms with ten or more staff now use AI, up from 12% at the end of 2023. Only one in ten of them uses it extensively, and lack of expertise is one of the most cited barriers. The gap is not the tools. It's the build.
ONS, Artificial intelligence in UK businesses: 2023 to 2026
Most technology advisors have either the technical depth or the commercial experience. Rarely both. I've built production AI systems, led national-scale transformation, founded four companies, and exited all of them.
I build the agents myself, alongside your team, in your systems, with your data. The deliverable is a process running in production with measured KPIs, not a recommendation to build one.
Four exits. VC-backed. Built to profitability. I've sat in every seat at the table, not just the technical slice.
The agents take the reading, matching and chasing. Your team keeps the judgement. You absorb more volume without hiring ahead of it, and every month you see exactly what the agents handled and what they cost.
Operated across AWS, Azure, GCP, and open-source stacks. The recommendation is always what fits your situation and budget. Not what earns a referral fee.
The short version of how I read this moment. Each one has the longer argument behind it.
The agent is now the easy part. Your records, your rules and who checks the work decide whether it helps.
The argumentAI writes the code now. The engineering was always the easy part. Leadership is what decides outcomes.
The argumentAgile is now the bottleneck. Stop planning the journey. Set a destination and navigate to it daily.
The argumentIf your AI strategy is just ChatGPT, you don't have one. The model is the commodity. The advantage is what you wrap around it.
The argumentMost failed AI pilots are recoverable. But don't let the people who built it be the ones to fix it.
The argumentThe quality you can't see is the one that kills you. AI is brilliant at the quality you can see, and weakest at the quality you can't.
The argumentDifferent sectors. One standard: measurable outcomes from systems running in production.
Agentic AI in production at a European insurance brokerage. End-to-end case resolution, full audit trail, deployed under regulator-grade scrutiny.
Four companies built from zero, all exited. Pattern-level lessons about taking technology from PhD bench to acquisition.
Chief Product & Technology Officer at NHS Wales. Found the savings the operating model couldn't see, and the structural shape that would keep them.
Start with a fixed-fee Process Audit, put your first process on agents in weeks, then keep it running and add the next one. Fractional CTO and board advisory when you need senior leadership as well. Built for companies of 50 to 500 people.
Pick one to three processes. I measure the baseline, show what an agent could run end to end, rank them by return, and write the build spec. Half the fee is credited against a build.
Deep review of your stack, architecture, AI approach, and team. Delivered as an investor-ready technical assessment. The clearest way to know where you stand.
Always-on agents that run one process end to end, on the systems you already have. Shadow mode first, then 10% of live volume, then scale on evidence. A person on every decision that matters.
Monitoring, exception review, accuracy and cost tuning, model upgrades and small changes, with a monthly report of work handled, hours returned and cost per item.
Architecture decisions, hiring guidance, investor updates, AI roadmap, and board presence. Senior technical leadership without the full-time hire.
Quarterly technology reviews, investor call prep, and a senior external voice for board credibility. An independent technical perspective when the stakes are high.
Engineering leaders, legal counsel, and government clients who’ve worked with me directly.
“Rafal was everything an expert should be: enthusiastic, immensely knowledgeable, personable and totally reliable. I have no hesitation in giving him my unreserved recommendation.”
“Rafal swiftly transformed the team dynamics, effectively improving previously low morale. His unique ability to win over his team’s hearts and minds was amazing to watch.”
“Technically he is brilliant. He is able to balance the technology and the user needs very easily. Quality driven. Not satisfied until the product reaches the level of quality he expects.”
The whole picture in one place: what to build, who should lead it, how to keep it good, and how to avoid the expensive mistakes.
Every working day you choose between doing the minimum and pushing a little further. Both become habits, and over years they compo…
Most late B2B payments can be recovered: the invoices are stuck, not refused. How an AI agent runs credit control, cuts DSO and le…
How an AI agent runs insurance claims intake for MGAs, TPAs and loss adjusters, why a named handler still decides every claim, and…
Tell me what you're working on. I'll respond personally. If there's a fit, we'll take it from there.
