Projects
Systems and sites currently live, with a link to each one. Private client work is listed without a link.
Multi-agent local news pipeline
A multi-agent local news pipeline for city editions with fact-checking and quality gates as release conditions, AI disclosure, and automatic unpublishing when a gate fails.
Evolutionary AI idea engine
Tracks live market signals, generates candidate ventures, stress tests each one adversarially and re-scores every idea daily, with a financial model, legal screen and investor memo. Re-scoring caught inflated scores, including one that fell from 91 to 61.
Hyperlocal weather to things to do funnel
A shared city platform combining live weather, forecasts, air quality and activity guidance with continuously checked local event listings.
Regional editions
City editions across Australia including Sydney, Melbourne, Brisbane, Perth, Adelaide, Canberra and the regional centres, each with its own masthead, local imagery and city weather.
Global editions
International editions including London, New York, Singapore, Tokyo, Paris, Berlin and Dubai, each localised down to institutions, courts and timezone so nothing reads as imported from another market.
Portfolio and case studies
This site: a static, server-rendered portfolio with case studies, a writing section and a hiring page, built on TanStack Start and deployed on Cloudflare.
Lead generation and nurturing with a built-in CRM
Lead generation and nurturing with a built-in CRM, first deployed on a live 254-apartment, A$292M residential development. Agents find and score buyer, investor and B2B leads, run personalised outreach, and nurture each lead by email, SMS, Instagram, Facebook and WhatsApp. A voice agent returns calls, DocuSign and Square close the sale, and human approval gates consequential actions. In production: 12,150 contacts, 8,950+ emails, 501 calls transcribed and analysed, 29 second median first response and 35 sales. A white-label edition for other developers runs on a separate database with fictional demo data.
Personal AI layer, beta
An invite-only personal AI layer people can open on any device. It picks the right model for each task and keeps working if a provider fails, credits at least three independent publishers in every answer and keeps customer-owned memory. Its agent builder turns one sentence into a working agent with success tests; every run is scored and weak agents are improved automatically.
Verified issue triage and guarded delivery
Triages issues across my platforms and routes each fix to the most suitable AI agent or to me, with verification gates and automatic rollback. In twelve weeks: 1,764 issues resolved and 1,257 merged pull requests.
Internal platform component
A routing layer that keeps generation running through a single-vendor outage, matches each task to a model that meets its quality bar, and caps daily spend.