My dissertation argued that AI would democratise access to professional tools — that capabilities previously locked behind expensive software and specialist knowledge would become available to everyone. Two years later, I am building exactly that thesis into products.
Invormed and Nutripedia are both AI apps designed to give normal people access to tools and insights that used to require professionals. One is for investing. The other is for health. They are at different stages, but they share the same core belief: AI makes it possible to build things that were previously only viable at enterprise scale.
Invormed: Holistic Investment Management
The investment industry sells two things: data access and advice. Bloomberg terminals cost $25,000 a year. Financial advisors charge percentage fees on assets. Retail investors get free apps with basic charts and no real intelligence.
Invormed sits in the gap. It is an AI-native portfolio manager that connects to your actual brokerage data and lets you have intelligent conversations about your investments. Ask "how exposed am I to emerging markets?" and it checks your real positions. Ask "should I rebalance?" and it compares your holdings against your own target allocations and investment theses.
The key insight: the value is not in the data display (any tracker does that), it is in making that data available to AI that can reason about it in the context of your personal research and goals. An MCP server gives Claude live access to your portfolio. A dashboard provides the visual layer. But the product is the experience of having a financial reasoning partner that knows your actual situation.
I wrote a separate post on Invormed — this is the overview of where it fits in the bigger picture.
Current status: MCP server and dashboard are live. Broker sync, AI-powered document import, multi-account management all working. Next: mobile interface for on-the-go investment queries.
Nutripedia: Evidence-Based Supplements
The supplement industry has a trust problem. There are thousands of products, endless conflicting advice from influencers, unregulated labels, and no way to know if what you are taking is actually doing anything. Examine.com does research but does not sell products. Labdoor grades products but does not track outcomes. Whoop tracks your body but knows nothing about what you are putting into it.
Nutripedia is an "honest broker" that sits in the middle. Three pillars:
The Research-Led Shop. Every product page shows AI-generated summaries of clinical trials, explicitly highlighting where the evidence conflicts. Only brands that pass a verification audit get listed. An AI assistant answers questions grounded in curated research papers — not marketing copy.
The Stack Manager. A mobile app where you log what you take daily, sync health data from wearables, and get monthly reports: "On days you took Vitamin D, your morning energy score was 15% higher." The correlation engine is the retention hook — it turns supplement-taking from guesswork into a feedback loop.
Own-Brand Products. Phase 3. Use marketplace sales data to decide which products to launch under our own label with full supply chain transparency.
The AI angle: research synthesis at scale. No human team can read every clinical trial on every supplement and keep summaries current. AI can. The same embedding and semantic search patterns I built for my personal knowledge system apply directly — except the corpus is medical literature instead of personal notes.
Current status: Phase 0. Waitlist is live, architecture is designed, research is done. Waiting for bandwidth — Rheos is the priority. But when I pick it up, the path to MVP is clear: seed the research assistant with papers, build the evidence interface for a handful of popular supplements, prove demand with zero inventory risk.
The Common Thread
Both apps take the same approach: connect AI to domain-specific data, and use it to give individuals access to reasoning that previously required expensive professionals or enterprise tools.
For investing, that means your AI assistant understanding your actual portfolio and research, not just answering generic questions about the stock market. For supplements, it means having access to the same body of clinical evidence that a nutritional scientist would reference, surfaced in a way that is useful without a PhD.
The shared tech stack across all my projects means context-switching between them is cheap. Patterns I build in one place — embeddings, semantic search, AI orchestration, MCP integrations — transfer directly to the next. These are not distractions from the main work. They are R&D for it, and they are the dissertation thesis made real.