AI for legal practices

Less time hunting through files. More time analysing them.

Lawyers, notaries and in-house counsel: an AI that prepares your files, finds the information in your sources and produces first drafts — under your rules for confidentiality and approval.

Tell us about your practice

case · preparation before reviewExample

Case 2026-118

42 documents received · 3 sources

type: commercial litigation
deadline: submissions to file
latest filing: opposing party's exhibits, last week

vérifié : Main contract and riders
vérifié : Correspondence between the parties
à vérifier : Exhibit IV cited but missing

AgentPuts the case in order before you open it

timeline rebuilt from the documents
clauses and deadlines spotted, with their source
missing document flagged, not filled in

HumainReview by the professional

the analysis, the strategy and the filing stay with the firm

Legal professions handle highly sensitive data. Using consumer AI services raises real professional-conduct risks: leaked information, unlogged processing, loss of control over case data.

In practice, AI can read, summarise and extract the information in a contract, draft a letter, watch a deadline or file documents. It replaces neither judgement nor the trust relationship with your clients: it prepares the work, you approve it.

We build these systems end to end for French practices, with confidentiality and professional conduct as the starting constraints: scoping, choice of a local, hybrid or cloud architecture, and documentation of the flows and the controls. Tell us about your practice in a first conversation.

Where AI steps in

Inside the life of a file, never beside it

An agent does not replace a step: it prepares the next one and hands back. Here is where this is worth it, and where it is not.

New file

Opening the file
documents received, filing
Agent
Files the documents and reconstructs the timeline
missing documents flagged
Humain
The associate completes the file
Research
sources the practice relies on
Agent
Queries the sources you approve and cites its references
sourced summary, reference by reference
Humain
The professional checks every source
Drafting
deeds, submissions, letters
Agent
Produces a first draft from your templates
a draft, never a deed
Humain
The professional rewrites and signs
Advice and hearings
strategy, advocacy, the relationship
No agent here, and that is not a gap: it is the profession, and it cannot be delegated.

What it changes, role by role

Partner or practice head: You want to commit the practice to an AI approach without exposing its reputation: we frame the scope, the professional-conduct obligations and the controls with you before any rollout.
Office manager or operations lead: You run the practice's secretariat, deadlines and output: AI can prepare files, standardise letters and watch deadlines, under your teams' supervision.
Lawyer or notary: You spend time on the first read of contracts and on document research: AI can prepare that work, while you keep approval and the final call.
In-house counsel or associate: You handle large volumes of documents: AI extracts, files and summarises the useful information, within the scope and rules you set.

Three scenarios

Named the way you name them

Nobody goes looking for “document analysis”. They go looking for a file that already stands on its own.

Scenario 01

Opening a file

Documents arrive, in batches and through several channels
The agent files them and reconstructs the timelinethe agent takes over
It spots the clauses and the deadlines, with their source
It flags what is missing rather than filling the gap
The professional opens a file that is already in orderyou take back the reins

Today: The first read mainly serves to find out what the file contains.
With an agent: A timeline and a list of gaps ready before the first read, which becomes a read for analysis.

Scenario 02

Preparing research

A question comes up on a specific point
The agent queries the sources the practice has approvedthe agent takes over
It reports the references found, with their citation
It writes a summary that links back to every source
The professional checks, discards and decidesyou take back the reins

Today: Research gets redone for every file, and last time's work is lost.
With an agent: A sourced summary to check, that states where every claim comes from.

Scenario 03

Keeping deadlines on track

Deadlines live in the documents, the diaries and people's heads
The agent picks out the dates cited in the file's documentsthe agent takes over
It cross-checks them against the practice's calendar
It flags at-risk deadlines before they close in
The team confirms, corrects or reschedulesyou take back the reins

Today: Tracking lives in a table that someone has to remember to reopen.
With an agent: A record kept from the documents themselves, and a warning before the deadline becomes urgent.

The other tasks we keep seeing

Automation angles worth exploring with your teams, under human supervision. To dig deeper, read our article : No-code AI: the quiet revolution in law firms.

Document analysis: Reading, summarising and extracting key information from your contracts, submissions and case documents, with structured, sourced output.
Research & case law: Support for document research and for extracting relevant precedents, under your control and from the sources you choose.
Drafting support: First drafts of contracts, letters and standard deeds, from your own templates, which you approve and refine.
Legal watch: Agents that monitor and summarise the regulatory changes that concern your field, at the frequency you choose.
Administrative processing: Automating data entry, filing and the production of standard documents — time-consuming work with little added value.
Deadline and time-limit management: Preparing case-tracking tables, reminding of procedural deadlines and flagging at-risk time limits before your teams approve.
File preparation: Gathering and putting a file's documents in order, spotting missing documents and a preparatory summary for the lawyer or notary.
Architecture and confidentiality: We can design a local, hybrid or cloud architecture, then document the data flows and the services chosen to fit the practice's requirements.

What bounds the agent

A practice agent adds nothing to the file.

Two tables, and they read together: what the agent is allowed to do before it starts, and what it did once it has finished.

Before — what is allowed

Each tool in the practice gets the rights the task needs, and not one more.

Example permissions granted to an agent, tool by tool.
Case management	Practice's messaging	Document sources
allowed : read a document	allowed : read	allowed : query the ones you approve
allowed : log a deadline	allowed : draft a letter	allowed : cite its source
denied : close a file	denied : send on its own	denied : add a document to the file

Example. The actual split is decided during framing, tool by tool, and access can be revoked.

After — what it did

Every document consulted can be traced back, to the second.

14:03:1242 case documents read
14:03:15timeline reconstructed
14:03:18exhibit IV cited, missing from the file
14:03:193 clauses flagged with their source
14:03:19file handed back for review
awaiting the professional's review

An example. The agent did not fill in the missing exhibit: producing an absent document would be invention, and it is set up to flag rather than fill in.

Your first agent

Where a practice would start

A first scope is chosen small: one type of file, its documents, a measurement after thirty days.

first project · file preparationExample

Document preparation for one type of file

What the agent covers

vérifié : Filing of documents received
vérifié : Timeline reconstructed
vérifié : Clauses and deadlines spotted, with their source
vérifié : Missing documents flagged
exclu : Drafting a deed

What it needs from you

ten files already handled
your filing plan
the sources the practice relies on

What stays with you

analysis and strategy
checking every source
drafting and signing

What we look at after thirty days: the time between receiving documents and the first useful read, and the number of gaps caught before the deadline.

estimate · files per monthYour figures

Files prepared per month

files / month

Average time to prepare one file

minutes

Time consumed today

≈ 30 hper month

This total is yours: it only comes from the two figures alongside. Only part of this time can be automated — we look at the volume, how stable the process is, the exceptions and the data available before saying so.

Get this process reviewed

How a project runs

Four steps, from framing to measurement

We move in stages, to check the value and the compliance before investing in a rollout.

01
Frame
- The tasks to support, the data involved, your professional-conduct obligations and the approval rules. You leave with an explicit scope and explicit success criteria.
02
Test
- A first assistant on the chosen case, on real practice data and within a bounded pilot. You adjust its behaviour with us.
03
Deploy
- Integration with your tools, the controls, the documented data flows and human approval on any binding action.
04
Measure
- The gap against the success criteria, reviewed with your team. You decide on scope extensions, one step at a time.

Frequently asked questions

Conduct, data and rollout

Straight answers on what AI can do inside a legal practice, and under what conditions.

Is AI compatible with professional secrecy?: Yes, provided the architecture is defined before any use. We build systems that limit data flows, choose the hosting with you (local, hybrid or cloud) and document every transfer. No service is used without your approval, and any binding action stays subject to human approval.
Is this use case really worth automating?: Not always, and we say so when it is not. We check the volume, the time it consumes, how stable the process is, how many exceptions it has and what data exists before proposing an agent. A rare or unstable task, or one full of special cases, costs more to automate than to do.
Can the practice's data stay in-house?: Yes. Depending on how sensitive the files are and on your obligations, the architecture can be local, on the practice's own servers or a host you choose, hybrid, or cloud. The choice is made during framing and every data flow is documented.
Does AI draft deeds and submissions in our place?: No. It produces first drafts from your templates and your instructions, which the professional reads, corrects and approves. Responsibility for the deed and the decision stay entirely in the practice's hands.
Do we need to train our staff?: The tools we deploy are designed to be simple to pick up, with no technical skill required. We also offer dedicated training so your teams understand how the system works, its limits and its control points.
How do we stop AI from leaking confidential information?: Through three combined mechanisms: minimal permissions (the system only consults what it needs), data flows documented and approved by you, and a log of every action. The prototype exists precisely to check these points before the scope is widened.

Let's talk about transforming your practice

A free first conversation to look at your professional-conduct obligations and the tasks AI could support.

Book a call

To go further

Other formats complement this offer, depending on where you're starting from.

Less time hunting through files. More time analysing them.

Inside the life of a file, never beside it

What it changes, role by role

Named the way you name them

Opening a file

Preparing research

Keeping deadlines on track

The other tasks we keep seeing

A practice agent adds nothing to the file.

Before — what is allowed

After — what it did

Where a practice would start

Four steps, from framing to measurement

Frame

Test

Deploy

Measure

Conduct, data and rollout

Let's talk about transforming your practice

To go further

Bespoke AI agent development

Workshop: frame and prototype an AI agent

AI for accounting firms

AI agent training