Development
Alongside selling cybersecurity and applied AI, I design and build the software: applications, agent tooling, and the integrations that connect them to real systems.
It is the same discipline that keeps my technical conversations honest. Working code has opinions about what is genuinely hard, what is a demo, and what will survive a security review.
Shipped
iPhone and Android apps built end to end with A2A agent tooling — agents handling research, engineering, media and review, with human decision-making, testing and App Store submission. Yard Yield, Coffee Quest and CradleSync are live on the App Store; Coffee Quest is also available on Google Play.
Shared baby care, in sync.
One shared care record for everyone who looks after your baby. Log feeds, sleep, nappies and medicines, hand over with confidence, and know what happened while you were away.
Grow what works, where you live.
An Australian edible-garden companion that turns your postcode, growing spaces and goals into practical seasonal guidance.
Visit yardyield.app →
Previously Home Harvest. Existing gardens and subscriptions continue in the same app.
Hidden gems & quirky finds. Never chains. Never landmarks.
For the cafes you’d walk straight past. Find character-rich independent cafes near you, save your favourites, and build a Grand Quest for a coffee walk worth taking.
In production use
One authoritative home for people, agents and the work they deliver.
Flow is a self-hosted project control plane for a person/organisation and the AI agent/s that help them plan and deliver work. It combines a browser workroom for the owner with a structured API and Model Context Protocol interface for agents. Both use the same project records, so a task, decision, comment, claim, dependency, approval, and completion outcome has one authoritative home.
The product is designed for controlled delegation. Agents can read project context, claim authorised work, record progress, request decisions, and attach evidence. The owner controls scope, approvals, and final acceptance. Flow records what each actor did and prevents a code merge, a passing check, or an agent assertion from silently becoming proof that the work is complete.
Flow is intentionally a focused, personalised system rather than a broad enterprise suite. Its value comes from connecting project planning, human decisions, agent execution, source control evidence, and an auditable completion record without splitting them across unrelated tools.
Flow is in production use and is used to plan and deliver the apps shown on this page.
More products
Designed and built the same way — AI agents working over A2A, with the decisions, testing and release kept human.
Your photos. Your pace.
A calmer way to work through a photo library: one image at a time, side-by-side comparison of near-duplicates, and nothing leaves your control until you decide it should.
Take a break. Stay connected.
Timed breaks from the social apps you choose, for iPhone and Android — pick the apps, pick 15 minutes to an hour, tap Disconnect. Phone and Messages stay available, and family controls are in development.
What I build
End-to-end builds — interface, data model and deployment — taken far enough to be used daily rather than demonstrated once.
The connective layer: MCP tooling, agents that discover one another and delegate work, and the integrations that let them act on real systems safely.
Software that runs on infrastructure you own, with retrieval over a governed knowledge store and no dependency on data leaving the network.
How it is built
The development work runs on the private Apple Silicon fleet — local language models, an Agent 2 Agent architecture, and a governed knowledge vault. Nothing leaves the network.
Buyers in regulated sectors are being asked to evaluate agentic systems, data residency and model governance. Having built these things, I can tell an architect what is actually involved — and tell a procurement lead what it costs to run.
