Watch a great GP work and you notice something the software never captured: they hold the whole fund in their head as a web. This founder was introduced by that LP's ex-partner. This new deal looks a lot like one they passed on two funds ago. The allocator who went quiet is the same person who asked, in a meeting last spring, for exactly the kind of co-invest that just landed in the pipeline. None of that lives in a column. It lives in the connections between things — and until now, the software stored the things and threw the connections away.
FundOS kept deals in one place, LPs in another, contacts in a third, documents in a fourth. Each was a tidy list. Ask it a question a GP asks reflexively — what do we already know that touches this? — and there was no answer, because the product had no idea the lists were connected. The same person showed up as a deal contact, an LP contact, and an introducer, as three unrelated rows. We fixed that. It's called the Fund Graph.
One person, not five records
Everything starts with people, because people are where a fund's network actually lives. The Fund Graph unifies every human you know — deal contacts, LP contacts, introducers — into a single identity, matched by email, conservatively. When two records clearly belong to the same person, they merge into one. When it isn't sure — two “John Smith”s, no shared email — it doesn't guess; it sets them aside for a human to confirm. The result is a page per person that shows every hat they wear at once: the deals they're on, the LPs they represent, who they introduced you to.
That single change is the keystone. Once a person is one node instead of five, every connection they carry snaps into place behind them.
“What connects to this?” — answered on the page you're already on
Open any deal, LP, or person and you'll find a Connections panel: everything linked to it, grouped and one click away — the rooms, the transactions, the meetings, the people, the fund it sits in. It's the answer to the question that used to require someone who'd been at the firm long enough to remember. And when this entity's name turns up somewhere it isn't formally linked — in a meeting note, in a deal memo — the graph quietly suggests the connection, shows you the exact sentence it found, and waits for you to confirm or dismiss it. Nothing is treated as true until a person says so.
You can also just draw a line yourself. See two things that belong together, link them, and they're connected everywhere that matters from then on.
A memory for every name, not just your LPs
We already had a “living brief” for LP relationships — a short, always-current summary of where things stand, what they've asked for, what they've committed to, synthesized from real history. The Fund Graph extends that memory to deals, portfolio companies, and people. A deal now carries the facts that matter about it, pulled from the meetings and notes around it and confirmed by a human before they count. A company remembers its own milestones. A person remembers the rooms they've been in.
The connections aren't a second copy of your data that drifts out of date. They're derived live from the records you already keep — so the graph is always exactly as current as the fund itself.
And then you can see the whole thing
The Fund Graph has a face: a full-screen, living map of your entire fund. Deals, LPs, people, rooms, companies — every node color-coded, sized by how connected it is, laid out by the gravity between them. Hover any node and its neighborhood lights up while the rest fades back. Click to inspect it; double-click to jump to its page. Focus on one entity and the map redraws around it, and your browser's back button retraces the path you took to get there, the way you'd retrace a train of thought. Search for a name and the view flies straight to it.
It's the part everyone reaches for first in a demo. But the map is the least of it — the value is that the connections underneath it are real, and that they're available everywhere, not just in the pretty view.
Built so the agents can read it too
Here's the part that matters most, and the part a knowledge graph bolted onto a note-taking app can't do. Every connection in the Fund Graph is readable by the AI agents working your book. When an agent drafts an LP update, screens a new deal, or answers a question in the copilot, it can traverse the same graph you can — who you know, how you know them, what connects this to that. It grounds its work in your fund's actual relationships instead of guessing from whatever happened to be on the page.
Two guarantees make that safe to rely on. Anything marked as material non-public information never leaves through the API or reaches an agent — the graph knows the difference between what a human can see and what a machine should. And the connections the software discovers on its own are always proposed, never assumed; a person confirms them before they're trusted. The graph is powerful precisely because it's disciplined about what counts as true.
A fund's real edge was never more data. It was attention, memory, and knowing how everything connects — the things the best partners carry in their heads and the rest of the team can never quite see. The Fund Graph makes that shared, durable, and legible to the agents working alongside you. Your fund always was a network. Now the software knows it.