Pre-launch · Base · no outside capital
We publish the method we run our own money on
Diligentia deploys its own treasury across DeFi lending markets, DEXes and yield venues. Every venue we touch carries a published risk score and a spending cap the contracts enforce.
Then we sell the method. Not a signal service and not a model you have to take on faith — a rubric with published weights that you can recompute from the same observations we scored.
Contracts in development. Nothing deployed, no capital deployed, no outside money accepted — and none will be.
Where the build is
- Method and threat modelScoring rubric, trust boundary, invariants, risk log
- Execution contractsRegistry, vault, allocator and the first adapter, tested against live Base venues
- Allocation engineThe optimiser and the data layer behind it. In progress
- External auditNot booked. Nothing carries capital before it reports
- First capital deployedOur own, and no date we are willing to invent
Updated as it changes, not as it is announced.
What the firm is
Two lines, one method between them
We deploy our own capital
Our treasury supplies liquidity to lending markets, DEXes and yield venues on Base. Where it goes is decided by the risk method, and bounded by limits the contracts enforce rather than limits we promise to respect.
We sell the method
The scoring rubric, the observation pipeline behind it and the venue history it draws on, licensed to protocols choosing integrations, funds allocating across venues, and treasuries doing what we do.
The second line is only worth anything because of the first. A risk method sold by people who do not use it is a spreadsheet with a sales team.
Why no deposits
Running our own book is a choice, not a stage
Most of what makes this sector hard to operate in comes from handling other people’s money. Investing your own is, in most places, not a regulated activity at all.
We would rather spend that budget on the method than on arguing a classification. So the firm carries its own risk, publishes what it finds, and is judged on a book anyone can check on chain rather than on assurances about one they cannot.
| Question | A protocol taking deposits | Diligentia |
|---|---|---|
| Whose money is at risk | Depositors', with the operator taking a fee either way | Ours. A bad call costs us before it costs anyone else |
| What the incentive is | Grow assets under management | Be right about venues, because that is the whole return |
| What you can verify | A dashboard the operator renders | A book on chain, and a score you can recompute yourself |
| What happens in a bad month | Fee income continues | It shows up in our own results |
The method
A number you can recompute, not one we assert
Every venue gets a 0–100 score across six dimensions — contract maturity, who controls it, how prices are read, whether a position our size could exit, what backs the loans, and what has already gone wrong there. The weights are published. Given the same observations, you get the same score we did.
- Low 0–29allocate freely, within the cap
- Elevated 30–59allocate, capped tighter
- High 60–100not whitelisted
Calibration profiles as a table
| Score | Band | Profile |
|---|---|---|
| 13 | Low | Mature lending market: immutable, 72h timelock, deeply liquid, no incidents |
| 38 | Elevated | Mid-tier market: upgradeable behind a 3-of-5 multisig, 24h timelock, two audits |
| 46 | Elevated | The same mid-tier market, with an uncompensated critical incident six months ago |
| 89 | High | Fresh fork: 14 days old, unaudited, single-key admin, TWAP oracle, thin liquidity |
How we operate
Four constraints the code enforces
Not a values statement. These are properties the contracts hold to, and every later decision has to fit inside them.
No outside money
We take no deposits and hold no third-party assets. That is a permanent choice, not a stage we are waiting to grow out of — it is what lets the firm be judged on its book rather than on its licence.
Limits the code enforces
Every venue carries a spending cap checked by the contracts on every move, not by the engine proposing it. A compromised allocation key misallocates within an approved set; it cannot reach anything else.
Adding risk is slow, removing it is instant
Approving a venue or raising a cap waits out a timelock. Cutting a cap, shutting a venue off and unwinding a position happen in one transaction with nothing to wait for.
Every number is read back
Nothing is reported from what we asked for. A transaction being accepted is not a fill, and a score we cannot recompute from stored observations is an opinion with a price on it.
Talk to us
What is worth a conversation now
Nothing is deployed yet, so the useful conversations today are about the method: whether it reads your market correctly, and what it would take to use it. We would rather be corrected before we are committed.
The domain is registered and the mailboxes are not live yet. Rather than print an address that would bounce, this says so.
Addresses appear here the day they work.