Why Nova AI Exists

Everything above — who we are, Jevons Paradox, Then vs. Now, The Bet — is the context. This section is the thesis. Why we’re doing this, for us and for the people we want to build with.

AI is changing everything — and it plays to our strengths

The shift is real and it’s happening now. Software engineering and product building — the things we’ve spent our whole careers doing — are being disrupted faster than any domain we’ve seen. Not “eventually.” Now. Between November and December 2025, Willison and Karpathy independently crossed the threshold. The next 18 months is the window.

This is an opportunity for us all because the disruption is precisely where our combined experience lives. Selling companies. Growing companies. Building companies. Legal structures, go-to-market, product validation, technical architecture — we have spent decades learning exactly the things that are being re-defined in real time. We should jump on this train.

Empowering others — their expertise + our AI scaffolding

Nova AI Ventures exists to empower other people — to take 20 to 30 years of their professional experience and help them transform it into something useful to humanity. Domain experts in wealth management, legal tech, marketing, medicine, education, sport, hospitality — people who have accumulated deep intuition about how their field works but don’t yet know how to leverage AI to build the next generation of products in it.

We provide the scaffolding: AI application architecture, legal and corporate framing, product-building methodology, go-to-market strategies, technology infrastructure. They provide the domain, the network, the credibility, and the founder energy. Their expertise + our AI-native stack = a category of venture that didn’t exist two years ago.

Services as the new software

Sequoia Capital recently published “Services as the New Software.” The thesis lines up exactly with where we’re placing our bets: the customer is buying the outcome directly. People want to pay for the outcome, not for a seat. The old SaaS model — per-seat licensing, feature-gated tiers, annual contracts — is being inverted in front of our eyes.

“Today’s judgement will become tomorrow’s intelligence. As AI systems accumulate proprietary data about what good judgement looks like in their domain, the frontier will shift.” — Sequoia Capital, Services as the New Software

Every product we build under Nova AI — SnapSell, Lumen, Otherth, My Story Machine, whatever comes next — must be pay-per-value at its core. Subscriptions include base token bundles. Overage is purchasable. Token margin is the primary revenue layer. Subscription is just the access gate.

How the internet itself is going to change

The interfaces we use to touch the internet are about to change fundamentally. People will not be scrolling, clicking, and filling in forms in five years. They will be talking to their agents. Those agents will live behind a Telegram conversation, or Ray-Ban glasses, or a voice assistant, or a browser extension — and they will connect to a constellation of APIs on the user’s behalf, paying for access with the user’s tokens.

The value will not be in the big LLM models themselves. The value will be in the constructed systems that run on those LLMs — the agent choreographies, the MCP tool integrations, the RAG pipelines, the domain-specific knowledge graphs, the token-metered APIs.

Why the Nova AI Stack is the bet

Every single product we build on the Nova AI Stack will be API-first, agent-to-agent ready, RAG-implemented from day one. Every activity collects structured data that becomes meaningful signal for the LLM agentic ecosystem. Every user interaction, every agent transaction, every API call becomes context the next product benefits from.

That’s why Nova AI Stack exists. Not to be a cool internal tool. To be the underlying platform that we implement in every single product venture — so that over the next 24 to 36 months we accumulate an ecosystem of interconnected, token-metered, agent-ready tools that cross-connect via MCP, A2A, and RAG.

The revenue thesis — API calls will outgrow app usage

Here’s the prediction, stated plainly: in the near future Nova AI will make more money through API calls than through direct application usage. Because the API is dedicated to agents. The agent connects to our API, checks the token balance, purchases tokens if needed (card on file, pre-authorized), does the job, reports back. The human never opens the app. The revenue still flows.

This is the future we are betting on: not just building products faster and more reliably, but building an interconnected ecosystem of AI-native products, linked by API, MCP, and CLI tools, with RAG and agent-ready intelligence at every layer.

Honest about the money — vision and revenue

This is why I believe Nova AI Ventures exists. This is why I believe we should be going in this direction. And there is a hard reality under all of it: we need to make money. We need to build products that generate revenue. Without the money, we don’t get to execute the vision. Money-making is the path, not the destination.

The rest of this blueprint — values, business model, Product Builder, Nova AI Stack, AI Management-Suite, founders’ roles, operator model, portfolio, revenue plan, cadence, governance — is the plan to get there. It’s a starting point. It’s a proposal. It’s what I want to discuss with the four of us so we’re aligned on both the future vision and the path to the revenue that funds it.

Agentic software engineering. An agentic internet where humans and their AI teams operate side by side. A Nova AI Ventures that makes really good money doing it. That’s the thesis. Money-making is the ultimate path toward achieving this all.