Case studies.

Challenge

A government entity needed an AI wellbeing platform that would meet DESC ISR v3 and UAE data-localisation rules — before a single line of it was built.

What I did

• Defined the solution and its technical requirements: multi-tenant architecture, AI service integration, content management, analytics, UAE data residency and a three-year cost of ownership.
• Ran focus groups with end users to define the user flows and product requirements.
• Compared vendors for the build, the voice-AI service and the AI chatbot — the chatbot was brought in-house after the comparative analysis.
• Set the data terms for the voice-AI vendor: hosting on a UAE cloud region, transcription inside the UAE, and no vendor access to user data.

Challenge

AI tools were arriving faster than anyone could check where they sent clinical data — and UAE health-data rules leave little room for guesswork.

What I did

• Reviewed AI tools before adoption across the group’s companies.
• Wrote policies and protocols that keep clinical data on UAE-hosted infrastructure, in line with ADHICS, DHA requirements and UAE health-data law.
• Ruled out transcription services that process clinical notes outside the UAE.
• Flagged a payment provider as unsuitable for clinical billing because its AI processing runs outside the UAE.

Challenge

A US retailer needed a complete point-of-sale solution across the country, on a fixed deadline.

What I did

• Designed and scoped the full solution: cash registers, receipt printers and barcode scanners for every checkout station.
• Coordinated several third-party manufacturers on one platform.
• Ran an accelerated client testing cycle to protect the deadline.

Challenge

Inventory picking and shipping across three large distribution centres needed to run on one integrated system.

What I did

• Designed and scoped the full solution: a warehouse management system, long-range handheld scanners, rugged computers mounted on forklifts and industrial label printers.
• Integrated it across all three sites, from inventory picking through to shipping.

Challenge

A healthcare manufacturer was opening a brand-new plant producing precision plastics, and needed tracking built in from day one — on the production line, at stock intake and at shipping.

What I did

• Closed, scoped and designed the full solution within about three months of first contact — the company’s first major project in Ireland.
• A warehouse management system with handheld computers and industrial label printers for stock intake and shipping.
• Specialised direct part marking (DPM) scanners to read tiny barcodes etched into the plastic parts as they moved through production.

Challenge

In large, busy labs — people constantly moving between tests — equipment kept going missing, and staff relied on memory to find it.

What I did

• Designed the solution: RFID-tagged lab and testing equipment, plus handheld RFID readers paired with a mobile app that shows how far away a specific item is.
• Launched it at the manufacturer’s European facility.
• Rolled the same solution out in Canada, at the Canadian facility’s request.

Challenge

Staff needed to raise an emergency instantly from anywhere in the facility, and reach every colleague at once.

What I did

• Built it myself with AI-assisted development: staff press SOS, choose an emergency code and location, and every other device sounds an alarm — even when locked.
• Found and fixed a privilege-escalation vulnerability, verified with a live penetration test; made alert records tamper-proof and moved alert dispatch server-side.
• Daily health checks and automated backups; every infrastructure account is owned by the organisation, not by me.

Challenge

Every purchase meant chasing three approvers by email, often bounced back and forth between them. Requests got lost, simple purchases took weeks, and there was no central record of what had been bought.

What I did

• Designed a standard Excel request template: product lines, quantities, prices and the reason for the purchase.
• Built the workflow in Teams and Power Automate — staff submit once, and the request moves automatically through clinical, finance and executive sign-off in turn.
• On final approval, accounts receive an automatic email with the request and product links, ready to purchase.
• Every completed purchase is logged in one central record: what was bought, when, and for which facility.

Challenge

A CV answers the questions I expect. I wanted visitors to ask the ones they actually have — without the answers drifting from the facts.

What I did

• The twin’s knowledge is generated from the same content file that renders this site, so the two can’t disagree.
• Guardrails keep it to what’s on the record: no invented figures, no client names, a pointer to email when it doesn’t know.
• Runs on a small open-weight model with streaming answers, input limits and rate limiting.