Law firms · Pricing

How much does an AI assistant cost for a small law firm?

Ours costs €1,000 to set up and €195 a month, with no lock-in. Here is what sits behind those two numbers, what genuinely varies, and how to compare providers who price on a different basis.

By Alejandro Maireles, CEO of Black Nodus AI · Published 25 August 2026 · Updated 9 September 2026

€1,000 setup (one-off, VAT not included) plus €195 per month of maintenance, with no lock-in. That is the whole price, it is published, and it is the same for a two-partner practice as for a twenty-lawyer firm. No cut-down edition, no per-user seat, no per-document meter, no extras that surface at month four.

Why we publish it at all

Most providers in this market do not. You get “request a quote”, and what comes back depends on how big your firm looked on the call. That is not necessarily dishonest — genuinely bespoke work is genuinely hard to price in advance — but it has a side effect worth naming: a price you have to ask for is a price you have to negotiate, and the firm that negotiates best is rarely the firm that needs the tool most.

We charge the same because the work is largely the same. Configuring an assistant for a firm that reviews leases and one that reviews share purchase agreements takes comparable effort. Pricing it by how much you appear able to pay would mean charging for the negotiation, not for the work.

What the €1,000 buys

The setup is the part that turns a general-purpose model into something that knows your firm:

The €195 a month covers hosting in the EU, the model usage, and keeping the thing working as your documents and your obligations change. It can be cancelled at any time, which is the only honest way to sell a monthly fee: if it stops being worth it, you stop paying.

What genuinely varies — and is not in our price

Three things move, and none of them are hidden fees. They are work that has to happen whoever you hire:

How to compare quotes that are not comparable

If you are looking at more than one provider, you will be handed numbers built on different bases. Four questions make them comparable:

If you are weighing a managed API against running a model on your own machines, Local LLM vs OpenAI covers that trade-off properly.

When you should not buy one at all

Worth saying plainly, because it is true more often than the industry admits. If your team asks a handful of repetitive questions a week, an assistant will not pay for itself — you will have bought a very sophisticated way to answer five questions. The same goes if the software you already pay for does part of this and nobody has switched the feature on.

The threshold is not a document count we can quote you. It is whether the same questions keep coming back and whether the answers already exist somewhere in your own files. If both are true, the maths works quickly. If neither is, wait.

What happens before you pay anything

We look at your documents and your workload first and tell you what can actually be automated — including, when it is the case, that it is not worth doing yet. That assessment is free and carries no obligation, and it is where the three variable costs above stop being estimates and become numbers.

Alejandro Maireles · CEO of Black Nodus AI
We build AI assistants that answer from your own documents and cite the source — private, EU-hosted, one price for everybody.

Not sure how it fits your business?

Book a free assessment: we review your case and tell you what can be automated and how much time you would save. No strings attached.

Book a free assessment →