Sized for you and your own data.

What you get

Built for work that has to be defended.

Custom analyses built to your question, and yours to keep. Rerun any of them on new data whenever the question comes round again, without rebuilding.

Every report carries the method it used and the code that ran, which is what makes it citable rather than merely persuasive. If a report is not right we diagnose it, fix it and resend at no cost.

Nothing is gated. A smaller plan is a smaller budget, never a worse answer.

How to start

Try it before price enters the decision.

Fourteen days, every depth open, no card. Bring your own data and your own question and find out whether the answers are any good. That is the only question worth settling first.

The current figures are published together on the pricing page and shown before anything runs. If you want to know why the plan is shaped this way, we have written that down too: how we price, and why.

Create your account → See the current pricing →

In detail

What the credits actually buy.

Plans run on credits. What a run costs moves with the model that reads your results and with what the work actually takes, so we do not quote a fixed rate. You are never charged for a run that fails, and we check you can cover one before it starts.

That is deliberate, and it is the part worth understanding. How far we go is a judgement about the question rather than a menu you buy from, so you ask the question you actually have instead of the one you can afford this month.

The second reason the work compounds: once an analysis exists it is yours, and re-running it on next month's data costs a fraction of building it. The first answer to a recurring question is the expensive one. Every answer after that is close to free.

Academic is one seat, and a seat carries a month of usage: enough to keep asking real questions of real data and to keep what you have already built up to date. If a month runs dry before the questions do, a boost buys that seat's month again at the same $20, for thirty days.

The demo works the same way at the same rates, which is the point of it. Fourteen days on your own data, spent only on runs. It is the whole platform at working scale, not a sample with the interesting parts removed.