Twenty years at Oracle, IBM, HP, Macquarie, Scotiabank, and Protocol Labs taught me what production looks like. I built AR Data to deliver the AI & blockchain systems most consultancies fake — bootstrapped, no investors, no prototype-and-disappear. The AR is me.
01 About
My arc started at Oracle and IBM, where I learned what enterprise delivery actually demands — not what the SOW says, but what it takes to ship a system that survives production. At HP/DXC I led large-scale infrastructure engagements across multiple continents.
At Macquarie Bank and Scotiabank I saw firsthand how financial institutions demand systems that cannot fail — and how few consultancies can actually deliver that. Then came Protocol Labs, where I contributed to IPFS and Filecoin infrastructure — the decentralized storage layer that powers Web3.
Most firms sell roadmaps and prototypes. Few can ship. The gap between a demo and a production system is where every engagement actually lives, and most consultancies never cross it. I started AR Data — one of the operating companies under ARCE Holdings Inc. — to close that gap. Bootstrapped. No investors. No bench warmers. Just engineers who have shipped, shipping again.
02 Holdings — The Group
Engagements run through the company that owns the domain — agentic systems through AR Data, vertical intelligence through Caiden, computer vision through SelfVision — but the engineering standard is the same across all of them. Selective joint ventures sit alongside when the right shape is a co-owned vehicle rather than a service engagement.
↳ One principal, one standard — engagements book into the operating company that owns the domain.
03 The Arc — Two Decades of Shipping
↳ Dates & locations approximate — tell me which to refine.
04 My Approach — The Loop
04 / 02 My Approach — WATED, Five Pillars
We name them WATED so the conversation with your team is the same conversation every time. Each pillar has its own choices, its own failure modes, and its own rollback story. We make all five explicit before we write a line of code — and we revisit them every time the system grows.
↳ WATED is AR Data’s own framework — proven across finance, enterprise, and Web3 engagements.
04 / 03 My Approach — BMAD, The Agent Crew
BMAD — the Breakthrough Method for Agile AI-Driven Development — is a public open-source method (BMad-Method) for AI-assisted software delivery. AR Data adopts it as-is and runs it inside the WATED frame, so the deliverables map cleanly to the five pillars.
Each hand-off is a typed artifact — replayable, reviewable, and reversible. Stories sized for a single agent to land. QA approves or routes back; nothing ships without a rollback.
05 Principles
Production is the only thing that counts.
If you can’t roll back, you can’t ship.
AI tooling moves fast. Maintenance doesn’t.
No bench. No prototype-and-disappear.
Agentic workflow. Human drives the outcome. AI supports.
06 Selected Work
A trading-floor compliance team buried in document review. Built a small graph of specialist agents with a rollback-by-default policy, a typed audit trail, and a human review gate on every write. Survived audit. Still in production.
Storage clients with petabyte-scale deal flow needed monitoring, retrieval verification, and remediation that didn’t require a human on every alert. Drawing on Protocol Labs work, we shipped an agent ops layer that handles the boring 95% — humans handle the rest.
Took a single-prompt support bot and decomposed it into specialist agents with typed tools, a router, and a golden eval set built from real historical tickets. Quality, cost, and latency moved the right direction at once. Rare combo.
If your team is between “we played with an LLM” and “we’re running it in production with our eyes open” — that’s the gap AR Data closes. Send a brief; let’s scope it.
07 Contact
AR Data takes a small number of engagements at a time. The discovery call is free, 30 minutes, and the fastest way to find out whether an agentic system is the right tool for your problem — or whether it isn’t.