Current Projects
I’m open to work and consulting on implementing powerful, effective AI for your organisation. No AI junk.
I build systems with AI that do real work. If you’re looking for someone who understands the engineering, the product design, and the human side of making AI actually useful, let’s talk.
lighterload
A personal relationship and family wellness coach, delivered through WhatsApp and Telegram. lighterload helps people carry less of the invisible mental load that comes with maintaining a household, remembering birthdays, keeping up with family, and staying on top of life admin.
The idea came from running OpenClaw — an open-source framework for building personal AI agents. After months of using my own AI agent to manage my life, I noticed it was genuinely good at the care-related stuff: remembering who I hadn’t called in a while, nudging me about upcoming birthdays, suggesting thoughtful gestures I wouldn’t have thought of. I realised that capability — that caring infrastructure — shouldn’t require setting up your own AI agent. It should just work, for anyone, on the messaging platform they already use.
What it does:
- Proactive care nudges — personalised check-ins that surface the things you’d otherwise forget: who you haven’t spoken to, upcoming milestones, life admin deadlines
- People memory — learns who matters to you and tracks the context of those relationships over time
- Age-appropriate parenting support — weekly check-ins grounded in evidence-based frameworks like Circle of Security and Emotion Coaching, tailored to your children’s developmental stages
- One-off reminders — “remind me to call Mum at 3pm” and it actually happens, reliably
- Calendar integration — forward emails with dates and events, get them extracted and added to your calendar automatically
- Couples plan — shared context between partners so both people have the full picture without duplicating effort
It’s not a general-purpose AI assistant. It does one thing — helps you be more present for the people who matter — and it does it well.
AgMap
AgMap is a digital network connecting farmers, researchers, and funders across Australian agriculture. The goal is to close the gap between what’s known in agricultural research and what’s practiced on farms — a gap that costs the industry billions annually in unrealised productivity.
The problem: Agricultural knowledge is fragmented across thousands of institutions, research bodies, and individual farms. A researcher in Queensland might solve a problem that a farmer in South Australia doesn’t know exists. A funder might invest in work that’s already been done. Everyone is operating in silos.
The scale:
- Australian agriculture is a $90B+ industry with over 85,000 farm businesses
- The sector invests $3B+ annually in R&D through organisations like GRDC, MLA, and CSIRO
- Despite this investment, research adoption rates remain stubbornly low — estimated at 20-30% for many innovations
- The platform maps relationships and knowledge flows across this entire ecosystem
AgMap is building the connective tissue that the industry is missing. It’s a startup I co-founded with my colleague Joe, and we’re working with some of Australia’s largest agricultural research organisations to make it happen.
Rapid Manuscript Editing
An AI-powered editing service for scientific researchers who publish in English as a second language. Researchers email their manuscript, and receive back a corrected version plus a detailed recommendations document — typically within two hours.
The problem: Millions of researchers worldwide need to publish in English-language journals to advance their careers and share their work. Professional human editing costs $500–1,400 per article and takes days. Many researchers — particularly in countries where the best AI tools aren’t directly accessible — are underserved by existing options.
How it works:
- Email-first architecture — researchers submit manuscripts via email with a prepaid token. No app to install, no website to navigate, no VPN required. Email works everywhere.
- Dual-AI quality pipeline — manuscripts are edited by one frontier AI model for language correction and structural recommendations, then independently reviewed by a second model from a different provider. Two different AI families cross-checking each other means fewer shared blind spots.
- What you get back — a corrected manuscript (.docx) with grammar, clarity, and flow improvements that preserve the original scientific meaning, plus a separate recommendations document covering structure, argumentation, and presentation.
- Token-based access — editing companies purchase tokens in bulk and distribute to their clients. A full admin dashboard tracks every job for transparent reconciliation.
The economics: At ~US$4 per page with API costs under $0.10, the service is roughly 10× cheaper than human editing while delivering results in hours instead of days. It’s positioned as a complement to premium human editing — a fast, affordable first pass that makes human editors’ time more valuable.
The opportunity: The global manuscript editing market is valued at $3–5 billion and growing fast. China alone has 1.87 million researchers — more than the United States — nearly all of whom must publish in English. The service is designed to scale through partnerships with established editing companies who already have the client relationships and local payment infrastructure.