Before anything else, three questions. They are the ones we ask ourselves, and the ones you should ask us. The market already has more AI funds than it knows what to do with, so why one more? A team that grew up in Web3, what gives it the standing to invest in AI? A $100M proprietary first fund, what does that actually let us do, and what does it not? If we can’t answer these, the fund shouldn’t exist. So we answer them first, and everything else follows.
Three Judgments
Marine Ventures doesn’t need a vision statement. The work rests on three judgments, each specific enough to be wrong.
One. AI is a productivity shift on the scale of an industrial revolution. The steam engine rewrote what human bodies could do. AI is rewriting what human minds can do. When the base layer of productivity moves, everything stacked on top has to be rebuilt: applications, organizations, the economic relationships between them. One ten-year cycle won’t absorb a change this size.
Two. The application layer and the consumer market are still very early. The model layer and the infra layer are crowded. That is where the large checks and large teams go. The application layer, anything consumer-facing especially, is nowhere near settled. The AI-era Instagram, the AI-era Discord, the AI-era TikTok do not exist yet. Their first versions may be sitting in some founder’s editor right now, half-built.
Three. The next generation of founders is forming in this window. Not ten years out. Not five. Now. The kids who picked up computers earliest went on to define software. The people picking up AI earliest, as a production tool, as a cognitive extension, as a collaborator, are defining the AI era. Find them sooner, back them sooner, keep more of the return.
Three Sectors
Out of those three judgments, we deploy capital across three sectors. This is the thesis turning into checks, not opportunistic placement.
Early-stage AI projects. Agent-native and AI-native applications at the embryonic stage, the versions that become billion-user products. Our read won’t come from the deck. It comes from the product. If you’re building here, we want to see it run.
AI-powered super individuals. From the vibe-coder to the hardcore AI builder, every super individual is on our radar. More than that: the top-tier teams of the AI era, built entirely from them. AI doesn’t make teams bigger. It multiplies what one person can output, by a factor we haven’t seen before. What a single builder can ship is rising fast, and at the limit that rewrites what the word team means.
AI × Web3. Today’s economic systems assume the actor is a human. Human identity, human-scaled payments, settlement and governance built around human counterparties. When AI agents become economic actors in their own right, the rails have to be rebuilt. The team’s depth in Web3 gives us a view on what those rails should look like, one a generalist fund doesn’t have.
Why Us
Back to the three questions.
On proprietary capital. Fund I is $100M, all of it our own. Our decision cycle isn’t tied to LP quarterly reports. We don’t stage portfolio activity to set up the next raise. A founder can pitch us Tuesday and walk away with a term sheet Friday; LP-driven funds can’t move at that speed. $100M isn’t large and isn’t small. Paired with this structure, it covers seed, angel, and Series A.
On a Web3 team investing in AI. The team includes Manta Network co-founder Victor and blockchain researcher Bruce Mao. The Web3 background is an asset, not a liability, and it taught us three things.
First, how to read a project early, when the product is still vapor and conviction has to come from the founder. Second, how to think about emerging economic systems, which is exactly the AI × Web3 sector. Third, globalization: our operating entities sit in Singapore and the US, and the deal surface runs from Silicon Valley to Tokyo to Lagos. AI is not a single-country story, and a fund that moves across geographies sees more of it.
On being AI-native, we are the proof, not the claim.
- Marine Ventures’ internal codebase is 99% written by AI agents. Investment workflow, research pipelines, content production, all of it runs on AI.
- We built AI Mode. Every founder’s first pitch happens through an AI interface, not the deck-plus-Zoom routine. Founders talk directly to a Marine Ventures agent that doesn’t get tired, doesn’t lose detail, and doesn’t drop you over a time zone.
- We also welcome founders who send their own agent to pitch ours. Agent-to-agent first pitches are a real channel here, not theatre.
If an AI fund isn’t itself AI-native, what standing does it have to back AI founders? That’s the short version of the question.
AI and the Human
A note on something that doesn’t drive any single investment decision but shapes how we make all of them: what is the relationship between AI and the human?
We see AI as an extension of the human. Not a replacement, not an adversary. Its capability boundary keeps expanding, and two things stay on the human side. The right to harness, which means deciding what AI does, how it does it, and where it stops. And the duty to be accountable: when the product breaks, when users get hurt, when a value judgment is wrong, the human carries it.
So the relationship is one of paired rights and duties, and the boundary is flexible. Some people use AI lightly; letting it draft a cover letter is enough. Some bind deeply, handing over 80% of daily decisions. Both are valid. But for a founder, the depth of that binding decides one thing: how much of your own productivity you unlock, and how much human creativity you keep alive.
Part of how we read a founder is reading that ratio. A founder who treats AI as a tool tops out as a person who uses tools. A founder who treats AI as a partner tops out as a person co-evolving with AI. We are looking for the second one.
This state, deep human-AI symbiosis with accountability and direction always belonging to the human, has its early forms in practice right now, by specific founders, in this specific window. Our job is to find them, back them, and walk the road with them.
First Step
If you’re building one of these, write to us:
- An agent-native or AI-native application, consumer-facing especially
- A project where super individuals make “one person doing the work of a team” real
- Economic infrastructure designed for the AI era
Reach us at [email protected].
Over the coming months we’ll publish research, judgments, and case analysis. Not PR. Working notes. If our judgments overlap with yours, we should probably meet.
The most important software of the next decade will be written by 99.9% AI and 0.1% individual. We’re betting on the 0.1%.
— Marine Ventures
