When a Managing Partner asks what a valuation change would do to the fund, the question lands with finance. Ad hoc scenario work sits outside what a fund administrator is normally asked for, so finance assembles it themselves. Level Ventures runs it in minutes instead.
Challenge
Level Ventures runs approximately $300 million in committed capital across a multi-entity structure of funds and special purpose vehicles. At that scale, the question that matters most is what a change in one position's valuation would do to fund-level NAV.
Answering that means assembling the fund commitment, the capital calls, the total called and current NAV before the model is built. Under the traditional model those inputs sit in separate places, each behind its own file, login or export. Finding the pieces is where the hours go, all spent before the question is answered once.
What Hanover Park deployed
Hanover Park's MCP collapses that gathering into a single read. Level Ventures' AI assistant queries commitments, capital calls and total called straight from the book of record. This works because the capital activity underneath is booked as it arrives rather than at month end.
Jonny Nulman, VP of Finance at Level Ventures, makes one request, in plain English and the assistant runs a hypothetical valuation against that data and calculates the hypothetical impact of specified valuation assumptions. He gets back the estimated fund-level impact, in a file he can work from.
Because the sourcing is instant, the question can be asked more than once. What if this position doubled? What if it tripled? What if it stays flat? His time goes into tweaking the fund analysis now, rather than hunting for inputs.
"A process that previously required significant manual data gathering can now be completed in minutes, giving us more time to review the output and underlying assumptions," said Jonny Nulman, VP of Finance at Level Ventures.
Recurring work, set up once
Jonny didn't stop at the one analysis. For all his subline documents he wrote a skill. It goes to Hanover Park, pulls the current unfunded figures, and hands back what the form needs, so the figures don't have to be re-gathered after every close and every call. He uses the same connection for treasury and operational reporting, where the figures come from one place instead of a separate file for each entity.
By his own account this is still early. Even so, the work that used to take a round of gathering now takes a single request. The skills he has written so far cover the work he repeats most, and the list is growing.
Results: the request cycle is gone
Level Ventures no longer files a request to answer a question about its own fund. It reads its own book of record instead, and because the underlying data is current, the answer is available the moment the question comes up.
About Level Ventures
Level Ventures is an investment firm focused on backing emerging venture capital managers and investing directly in select high-growth companies alongside its fund partners. The firm operates a fund-of-funds and co-investment platform designed to provide investors with diversified access to early-stage venture opportunities while maintaining a disciplined, research-driven approach to manager selection.
About Hanover Park
Hanover Park builds AI systems that run the financial operations of the world's most sophisticated investment firms. As an AI-enabled service and fund administrator, we've become the way fund CFOs adopt AI. We've raised $31 million from Emergence (investor in Salesforce/Zoom), Susa (investor in Robinhood/Flexport) and angel investor fund CFOs from the largest investment firms. Our mission is to be the financial infrastructure to orchestrate how the world's capital is operated, governed and allocated. Stripe for payments. Ramp for expenses. Hanover Park for investments.
