Who it’s for · Funders · Foundations
Decision-grade evidence across the whole portfolio, not only the grants big enough to evaluate.
Many foundations can justify bespoke impact evaluation for only a selected group of grants. Prova is being built to change that arithmetic: the same standard, extended across the portfolio, on a timeline that fits how you fund, from a method you can read and hand to your evaluation lead.
The bind
Credible evidence gets rationed to the grants that can carry its cost.
Bespoke impact evaluation can be difficult to justify across a whole portfolio. Much routine reporting therefore relies on outputs, self-report, or before-and-after change. Some studies conclude after the funding decision they were meant to inform. Evidence paid for by the funder carries an incentive risk that has to be managed. None of this is a failure of will. It is the cost of knowing, the constraint Prova is being built to move.
What Prova does
The same standard, extended across the portfolio.
Prova works from the records programs already keep. Frontier AI models can now perform much of the reading and synthesis work. What remains to be demonstrated is how far that capability lowers the total cost for grants that have never justified a traditional evaluation. The standard of evidence does not move. Prova is being built to establish how far the price of reaching it falls. The standards behind the work are written down: how Prova’s four-grade claim-evidence rubric is applied, how operational variation is tested before it is treated as a credible natural experiment, and how a design is stress-tested before it runs. Read them, or hand them to your evaluation lead.
This extends what your team can reach. It does not replace their judgment. The hardest calls, what an effect means in a place and what a portfolio should do about it, stay with the people who hold them. AI does the labor-intensive reading and synthesis under human supervision. How far that capability lowers the total cost is still being tested. The judgment stays human.
How to start
Start with one grant.
A natural place to begin is one grant that has never justified a bespoke impact evaluation: a bounded assessment of what its existing evidence can support. The timeline, work required, and cost depend on the records, question, governance requirements, and design. It is the cost-of-knowing thesis tested on one grant. From there, the same method extends to a live program worth following over time, and to a full causal study when a question warrants one. All to the same standard.
Start a conversation
Bring your hardest grant to measure.
If there is a corner of the portfolio you have never been able to see clearly, that is the conversation to have. The first one is a fit check, with no obligation.
Common questions
The questions a careful reader asks first.
Is cheaper evidence weaker evidence?
A lower price does not require a lower standard. Frontier AI models can now perform much of the reading, extraction, and synthesis in bounded tasks. Prova is being built to establish how far that capability lowers the total cost of reaching a given grade in its claim-evidence rubric without shifting work or risk elsewhere. Design labels do not guarantee credibility. A modest claim remains modest, and a thin record remains thin.
AI can be confidently wrong. Why trust it here?
It can, and that is a central design problem. AI does the labor-intensive reading and synthesis; people decide whether a finding holds, what it means here, and where the evidence runs out. Prova is being built so claims retain provenance and uncertainty. Its reliability across real engagements is still being tested.