AI for lawyers: what it really covers and where the risk begins

AI for lawyers: what it really covers and where the risk begins

A contract analysis that used to take a day is done in ten minutes — and the next thought arrives immediately: what if there had been an error in it. Below is both what an assistant actually produces and the verification protocol without which the result must not be taken into work.

A contract analysis that used to take a day is done in ten minutes — and the next thought arrives immediately: what if there had been an error in it. Below is both what an assistant actually produces and the verification protocol without which the result must not be taken into work.


Work that consists of rereading

A lawyer in a small company or in private practice spends most of the time not on difficult decisions but on repetitive reading.

A contract arrives from a counterparty. It has to be read in full, gaps found, wording that works against your side flagged, and comments assembled. A standard contract takes between an hour and half a day. There are several a week.

And the work itself is repetitive: the same places, the same risks, the same list of what templates usually omit.

From my audience survey — the words of someone who has already tried it: a contract analysis that took a day was done in ten minutes.

And the thought that follows chills you: what if there had been an error? Below is both what actually comes out and the protocol without which this cannot be used.

What the assistant produces

I took the Corporate lawyer from the catalogue and set one task with no documents attached:

Draw up a checklist for reviewing a services contract from the contractor's side. Which clauses are essential, which wordings work against the contractor, what is most often missing from standard templates.

The corporate lawyer — its specialisation and range of tasks

The assistant immediately set out a frame not every practising lawyer articulates: a paid services contract carries more risk specifically for the contractor — because of the client's unconditional right to withdraw unilaterally at any time, and the intangible nature of services themselves.

Then the analysis in three parts.

Part one: essential and protective clauses

The first part of the checklist — protective clauses with references to statute

Five clauses, each with a reference to the relevant provision and ready wording:

Subject matter and the brief. Anything not stated in the brief is an additional service and is paid for separately.

Automatic extension of deadlines. An explicit provision: if the client delays source data, access or approvals, the deadlines extend by the length of the delay plus several working days.

Automatic acceptance. A firm deadline for signing off, and if the client sends neither a signed acceptance nor a reasoned refusal, the services are deemed accepted without reservation.

The right to suspend work if payment is even a day late — with no penalty to the contractor.

Calculation of compensation on unilateral withdrawal. An itemisation of costs actually incurred: licence purchases, engaging subcontractors, reserving resources.

Part two: dangerous wordings

Red flags and what is missing from standard templates

This is the most useful part in practice. Five red flags, and all of them come up constantly:

Payment tied to external events — "after financing is received", "after regulatory approval". A service is paid for the performance of actions, not for somebody else's business events occurring.

"Rectifying comments until the client is fully satisfied" — endless free revisions. Comments are accepted only within the agreed brief.

Disproportionate penalties — 0.5–1% a day with no cap. The assistant also names the specific move: reduce it to 0.05–0.1% and introduce a cap on total liability at the value of the fee for that stage.

Transfer of rights to the deliverable before payment. Rights should pass strictly after payment in full.

A direct penalty for the client withdrawing from the contract — courts treat that as void, and what is needed instead is a detailed calculation of costs.

Part three: what templates leave out

Clauses almost never present in a contract downloaded from the internet:

The protocol without which this cannot be used

Now the honest part, and it matters more than everything above.

A model states non-existent provisions as confidently as real ones. An error looks exactly like a correct answer: a section number looks like a section number, a reference like a reference. There will be no signal saying "I am unsure here".

For a lawyer that means not "do not use it" but "do not use it without a process".

Separate the answer into two kinds of content. Verifiable statements — provision numbers, dates, sums, references, quotations. And reasoning — the logic, the structure, the wording. The second you judge with your own expertise. The first is always checked against the primary source, even when it looks obviously right.

Supply material rather than relying on memory. A request of "here is the contract, analyse it against these points" is more reliable than "tell me what contracts usually contain". When all the material comes from you, there is nothing to invent — and that is the safest mode of working.

Nothing goes out without a live lawyer's check. AI speeds up analysis and drafting. Responsibility for the wording lies with the person who signs it.

And clients' personal data does not go into a chat. If in doubt, treat it as forbidden.

What a lawyer can safely give to AI and what not

Almost always safe: analysing a supplied document, finding internal contradictions in it, comparing two versions, preparing checklists and questions for a counterparty, restating something complex in plain language for a client.

Requires full verification: any provisions, deadlines, details and anything that was not in the material supplied.

Never handed over at all: final decisions with legal consequences, and data that must not leave.

Two things people usually do not know

Assistants have checking modes. Every answer ends with an offer to assess the work and suggest improvements. There is a focus group mode where the model goes through its own answer with the recipient's eyes, and a refinement mode against assessment criteria. That turns a first draft into a checked document without your involvement.

Assistants can be put in a chain. The output of one goes into the input of the next. The corporate lawyer analysed the contract — the paralegal assembled a letter to the counterparty with the comments. What gets assembled is a process rather than an answer.

How a lawyer earns from this

First — speed for yourself. A contract analysed in ten minutes instead of a day. This is not about cutting staff; it is about the freed time going to work nobody previously got round to.

Second — as a service to businesses. A legal risk audit of a small company: contracts with clients and contractors, staff confidentiality undertakings, rights to work created by contractors, public documents. Work like that costs from $2 400 — that is the price by our methodology, not market statistics. The bulk of it is analysing existing documents and mapping risks, and that speeds up several times over.

Where to start

The "Legal department" category holds thirteen assistants: corporate, tax, employment, intellectual property, paralegal, insolvency, property, succession and others. In total the platform holds 200+ AI assistants across 14 categories.

Worth naming separately is the express course "Perplexity as an investigative tool" — it is not about law, it is about a method of verification: how to check a fact, a reference or a figure in minutes instead of taking a text on trust. For this profession it matters more than any other. It costs $9 and is included in Basic.

Registration is free and opens three days of full Basic access — all the assistants, the course on verification and five more Basic mini-courses.

Do one thing today: take a contract you are working under now and run it against the third part of the checklist — the one listing what templates leave out. The correspondence protocol, the cap on revisions, the criteria for a reasoned refusal. If even two of those are absent, you have found your open risks in five minutes.


The second approach: not a chat but a conveyor

Everything above is work in dialogue with an assistant. There is another format, and for volume it suits better.

Cascading prompts in spreadsheets

The “Earning with ready templates” course contains spreadsheets with an integrated model: a large task broken into micro-steps, each living in its own cell and taking the previous result as input. The formula can be dragged down a thousand rows — and a thousand tasks are processed at once. How this works step by step is covered separately: Thirty posts in one launch.

What is available for a lawyer:

For personnel documents (duties, KPIs, a job description) the HR Manager assistant fits — HR advice based on CHRO experience; the result still needs review. The “Earning with ready templates” course is included in the Basic plan.

When a chain needs images, video and sound

When a chain needs images, video or sound, a spreadsheet is not enough. You can hand such a task to Cerebrum: describe it in plain words, the right Gen AI models are picked automatically, you pay from your balance for what is actually done, and the price is visible before launch. For linking external services, use the n8n and Make automation templates (Creator and above).


Ready tools for documents

The “Automator” course (Plus plan) has the lessons “Working with files. Contract analysis” and “Working with files. Report analysis”: taking documents apart and finding contradictions between sections and discrepancies between text and tables.

For scans and photographs of documents there is the open-source tool Chandra — text and document structure recognition (the description is open to everyone, the repository link comes with Creator).

For a lawyer that is a way of working with scans and photographs of documents rather than only with text files.


Thirteen lawyers in the catalogue

The "Legal department" category in the assistant catalogue — thirteen roles, Basic plan. The entry threshold is zero: you open a chat and describe the situation in your own words.

The catalogue and descriptions are open to guests; the chat links come with the Basic plan.

Where this sells as a service. The combination of an assistant plus verification by protocol can be packaged into your own offer — implementing AI in a small company's legal department. Contract analysis step by step is shown in the lesson “Working with files. Contract analysis” of the “Automator” course (Plus plan).

Skills for document work

The catalogue covers exactly the part of the work this article calls safe: you supply the material, the tool works with it.

What matters more than the list. All of them work on one principle — you supply the document, they work with it. That is the very mode in which the model has nothing to invent, and the only one this article considers acceptable for a lawyer.

What to read next

How to verify a fact with AI search — the verification method that matters more than any other for this profession.

Rights to a face in advertising — an adjacent subject where contractors need a lawyer.

A legal shield for a business — the same skill sold as a service.

A company knowledge base as a service — what to do with scattered internal procedures.

Best open LLMs 2026 — the full guide to open models.

The shop did not deliver — a real case step by step: the letter, the penalty, the channels.


Also on AI SKILLS