Mayeshal charges a per-user subscription for the platform. Your AI consumption is billed to your firm directly by the provider, at published rates, with zero markup from us. Two numbers, both visible, and the amount we make does not go up when your usage does.
Every tier includes the whole platform. There is no feature tier. The differences are users, storage, connectors and how much of our team is on it.
Setup is a fixed price and a fixed scope. It covers provisioning, identity and single sign-on, connecting your systems, configuring the models and the agents, end-user and administrator training, and ninety days of hyper-care after launch. We do not bill hourly against it and we do not send a change order because a connector took longer than we thought.
Once the platform is live, a vertical pack installs in two more weeks.
We do not publish a return-on-investment calculator, because as of 2026 essentially nobody in legal can measure return on investment on AI, and a projection you cannot check is not worth your time. Here is what the work itself actually cost at a live firm.
Every figure here is verifiable against a Mayeshal product reference file.
Those are the real line items from reading a 127 GB corpus of medical records, bills, police reports and photographs. Each number counts a different thing, so here is the arithmetic stated plainly. 44,956 documents were catalogued and 331,486 pages were indexed. 37,852 of those documents were given an AI summary; the rest are files that no longer exist at the source, files that were never documents, or files with nothing to summarise. The $0.0046 figure is the cost per summary measured on a sampled run. The $156.55 is what the whole build actually booked: $149.16 for the 37,852 summaries, which works out at about $0.0039 each once the whole corpus is averaged, plus $7.39 for the vision pass over the photographs. Optical character recognition added nothing, because it runs on our own hardware.
We publish these because a partner can do the arithmetic on their own case volume in their head, on this page, without talking to us. That is the point.
Your own consumption will differ, because it depends on how much your team asks and how much of your corpus is scanned rather than digital. What will not differ is who you pay for it and what we add to it.
Plenty of good software is sold per seat, and Mayeshal is one of them. The thing worth being careful about is different: a subscription plus consumption credits, where the credits are priced by the vendor, the underlying model cost is not disclosed, and the definition of a credit belongs to the vendor. "Unlimited" in that structure is a word, not a term.
Across 2026 the whole category has been moving toward consumption pricing. Vendors have been public about token volumes rising sharply and about repricing to match. That movement is real and it is not unreasonable. The question for a buyer is simply whether they can see both halves of the bill.
With Mayeshal you can. The subscription is a published number on this page. The consumption is a bill your firm receives from your provider, at that provider's published rates, for accounts in your own name. We are not in the middle of it, and we do not want to be.
There is also a reason this matters more to a plaintiff firm than to an hourly one. Where a firm sells time, saving time is a complicated thing to buy. Where a firm works on contingency, the entire benefit lands on the cost side of the ledger, and cost per case is arithmetic your firm already does every week.
Thirty minutes, on a screen share, on a matter you pick. We connect nothing, install nothing and ask for no credentials to have that conversation. If it is not obviously useful in the first ten minutes, we will tell you so and you get your afternoon back.