I build AI products, advise the ventures around them, and research what comes next.
I'm Gregory O'Connor. I build the first useful version of an AI product, run the venture and infrastructure diligence behind big decisions, and advise teams through launch — keeping evaluation, privacy, cost, and handoff from becoming afterthoughts.
Most of the hands-on build work right now is with Claude, OpenAI, and Gemini, alongside coding agents, retrieval layers, structured outputs, and evals.
Best fit: founders, operators, and teams where an AI product — or the decision to back one — has to hold up under real scrutiny.
Recent work
A mix of recent engagements — products I've built, ventures I've helped evaluate, and harder problems I'm still pushing on. Details stay confidential; these describe the capabilities, not the proprietary work. Each links to a fuller write-up.
- AI applications and SaaS products: Scenefiend, a consumer SaaS taken from concept to a beta-ready release candidate — now in private beta. Read more →
- Fintech venture advisory: end-to-end venture diligence and go-to-market planning for a new consumer payments product. Read more →
- Startup and board advisory: board advisor to NAVIA AI, a SaaS company shipping an AI executive-coaching product now in beta. Read more →
- Secure, distributed AI infrastructure: control-plane architecture for permissioned distributed AI workloads — coordination, telemetry, governance, and operator visibility. Read more →
- Industrial and infrastructure advisory: AI-enabled analysis spanning engineering feasibility, commercial strategy, and geopolitical-macroeconomic scenario planning. Read more →
- Legal and public-interest AI: source-grounded document review and citation finding, plus private model training and self-hosted deployment that keeps confidential data in-house. Read more →
- Clinical and health knowledge systems: source-grounded synthesis of medical literature with explicit uncertainty and a clinician in the loop. Read more →
- Multilingual narrative monitoring: a human-reviewed pipeline for tracking messaging themes across languages and channels. Read more →
- Explainable-AI research: a novel mathematical and linguistic approach aimed at more human, more explainable intelligence. Read more →
I don't do political microtargeting, covert persuasion, or surveillance work. If your use case needs careful boundaries, we figure those out first.
Where this fits
The common thread is document-heavy work where teams need source-grounded answers, clear uncertainty, private data handling, and a human review path. The domain changes; the constraints repeat.
- Legal and public-interest workflows: AI-assisted citation finding, authority recommendations, and source-grounded document review that help legal teams deliver clearer guidance while keeping confidential data and human judgment protected.
- Clinical and health knowledge work: source-grounded synthesis of medical literature and guidelines, with explicit uncertainty and a clinician always in the loop.
- Real estate operations: portfolio intelligence, resident service, engineering and maintenance workflows, and diligence tools that connect leases, rent rolls, work orders, inspections, operating reports, communications, and human follow-up.
- Internal knowledge systems: assistants that work against private reports, policies, and institutional memory instead of the open web.
- Build-vs-buy decisions: helping teams decide what should become software, what should stay a process, and where AI actually changes the workflow.
How I work
Most projects start with one problem that matters, a pile of imperfect context, and too many options. I narrow the scope fast, get something testable in front of real users, and make quality, latency, cost, and failure modes visible before the system grows. If the first version looks promising, we harden it; if it doesn't, you find that out in weeks instead of after a long, expensive build — and throughout, I can translate between business leads, subject-matter experts, vendors, and engineers.
Prototype to release-candidate path
A useful first version is more than a demo that works on friendly inputs. I build the product path and the system together: scope, model behavior, retrieval, tests, deployment, and the handoff plan.
Document and knowledge systems
Knowledge systems live or die on boring decisions: what the retrieval layer returns, how sources are cited, and what happens when the model is wrong. I design that layer so quality problems show up before users find them.
Architecture, evals, and what to log
Once a system has more than one model or service, the hard part is rarely the model. It is what calls what, what gets logged, what needs human review, and what becomes an audit problem later.
Working alongside your team
I can build independently or join an existing product or engineering team for the AI-specific work. The useful version is staying close enough to ship, then handing off the prompts, eval set, docs, and cost assumptions cleanly.
How engagements work
Start with the smallest engagement that can answer the important question.
Strategy + Architecture Sprint
1–2 weeksI look at the workflow, the data, and the existing systems, then write a build plan with explicit go/no-go gates. Not a vision document.
Prototype Sprint
2–4 weeksA working version in your environment, narrow enough to prove value. Includes evaluation, cost assumptions, and an honest read on whether to harden or stop.
Build + Ship
4–10+ weeksI stay with the build through integration, testing, and handoff. The goal is a system your team can operate without depending on hidden prompt tricks or heroics.
About
I'm Gregory O'Connor. I build AI systems that have to hold up outside the demo. The work I like sits where product judgment and model behavior stop being separate problems.
MIT-trained Ph.D., based in New York. I've been building technical systems for a long time. Recent work includes Scenefiend, a consumer SaaS now in private beta; venture advisory to an early-stage consumer fintech; and a board-advisor role at NAVIA AI.
The problems span entertainment, fintech, legal and public-interest, health, energy and industrial infrastructure, and AI infrastructure itself — different domains, the same discipline about data, evaluation, and where humans stay in charge.
I'm usually most useful when a team has a real workflow, partial data, and pressure to ship before the system is fully understood.
Send the workflow
A first email does not need to be polished. The useful version says what the workflow is, who uses it, what data is involved, and what would make the first version worth keeping.