A small team has no board forcing hard questions every month. We built an AI investor that reads the numbers, grades last month's promises and says what we would rather not hear.
Nova is run by a small team. That has real advantages, but it also means nobody sits across the table once a month and asks the questions we would rather skip. So we built something to do it: George the VC, an AI investor that reads our numbers and runs a monthly board review with the founders.
Why we wanted a board that disagrees with us
A company with outside investors gets a board meeting whether it feels like one or not. The date is fixed, the deck has to be ready, and someone will ask why the plan from last quarter did not happen. A small, founder-led team has none of that by default. Weeks blur together, priorities shift quietly, and the hard conversations get pushed to a better moment that never comes.
We tried using general AI assistants for this and hit the opposite problem. Ask a chat assistant to review your month and it tends to agree with you. It finds the positive angle, softens the bad news and praises the effort. That is pleasant and close to useless.
What we wanted was one fixed day a month where something reads all the numbers and says, plainly and in writing, what we would rather not hear.
What George reads
Before each review, George gets the same set of inputs:
Revenue and the MRR bridge. Not one total, but how it moved: new revenue, expansion, contraction and churn, each on its own line.
The sales pipeline. What is in it, what moved forward, what stalled and what was lost.
Customer cohorts and retention. Whether customers who joined in a given month are still with us, and whether newer cohorts behave better or worse than older ones.
Cash and runway. What came in, what went out and how long the current plan lasts.
What engineering shipped. The actual releases of the month, compared with what we said we would ship.
The point of reading all five together is that each one can hide a problem in another. A strong sales month can mask a retention issue. A big release can distract from a pipeline that has gone quiet.
The promises ledger
Every review ends with a short list of promises for the next month. They are specific and written down: something we will ship, a deal we will close or drop, a number we will fix.
The next review starts by grading every one of those promises. Each gets one of five grades: kept, partly kept, open, dropped or broken. Dropped means we decided, on purpose, to stop. Broken means we said we would do it and did not, without a good reason.
This is the part that changed our behavior the most. In a normal month, a commitment that becomes inconvenient just fades away. Nobody mentions it again. With the ledger, nothing quietly disappears. If we want to stop doing something, we have to say so and accept the grade.
The verdict
Each review closes with a verdict, the way an investor would sum up a meeting. There are five, from Hard pass at the bottom to Preempt at the top, and each comes with its own portrait of George, from unimpressed to ready to write the check.
The verdict is deliberately blunt. It is not a score to optimize, but it is hard to ignore when a month you felt good about comes back with a verdict that says otherwise, along with the reasons.
What changed for us
The biggest change is not in any single decision. It is that uncomfortable things now get said out loud, on a fixed day, in writing. A worry that would have stayed vague for weeks becomes a line in the review and, often, a promise for next month.
It also made our planning more honest. Knowing that every promise will be graded, we make fewer of them and we make them more precise. And because the review reads the same inputs every month, trends that are easy to miss week to week become obvious when they are laid side by side.
It does not replace talking to customers, advisors or each other. It gives us a recurring moment where the business is looked at as a whole, by something that has no reason to be polite about it.
George is now a product for other founders
We turned George into its own product, so other founders can run the same review. George the VC runs the same kind of monthly board review for any founder who wants one: it reads your numbers, keeps your promises ledger and gives you a verdict.
Find answers to common questions about our platform
George the VC is an AI investor built by the Nova team. It reads a company's numbers once a month, asks the questions a board would ask, grades last month's promises and ends with a verdict.
Revenue and the MRR bridge (new, expansion, contraction and churn), the sales pipeline, customer cohorts and retention, cash and runway, and what engineering shipped during the month.
It is available on a fixed day every month, it reads every number, and it is set up to push back rather than agree. It does not replace human advisors, but it removes the excuse of not having a board meeting.
