Law firms pilot v0.1.0

Conclusions Drafter

Drafts 80% of a legal writing (court conclusions, memo, letter) from a short brief + the case-file exhibits. The lawyer arbitrates the 20% that needs judgement. Saves 3-5 hours per writing, ~200 writings/year/lawyer.

See the live demo Source code on GitHub

Measured gain

3-5 h

par écriture juridique

Across 200 writings/year/lawyer × 20 lawyers: 12,000-20,000 hours recovered, ~€3-5M of time freed at billable rate. The lawyer always remains signatory and decision-maker - the agent never signs.

The problem

Writing court conclusions is 30-50% of a litigation lawyer's time. And 80% of that writing is structurally repetitive.

Today

Rédaction manuelle

30-50 % du temps d'un avocat contentieux

4-6 h / conclusion (30 p.)

× 200 écritures / an / avocat

→ 800-1 200 h / an / avocat de rédaction pure

· Rappel des faits avec citation pièce par pièce
· Syllogisme juridique, citations de jurisprudence
· 80 % structurellement répétitif (style maison, dispositif)
· Le travail à valeur ajoutée = 20 % du temps, 100 % de la valeur

80 % du temps facturé sur du mécanique

With Conclusions Drafter

Brouillon à 80 %

L'avocat relit, arbitre, signe

1 h / conclusion

0 signature automatique

→ 12 000-20 000 h récupérées sur un cabinet de 20 avocats / an

› Style maison récupéré via search_style_house
› Jurisprudence sourcée via search_jurisprudence, jamais inventée
› Chaque inférence marquée [à valider], pièces citées par numéro
› Signature finale toujours « Maître [avocat - à compléter] »

L'avocat reste signataire et décisionnaire

How it works

Four steps, from incoming document to human decision.

1

Lawyer brief received

3-15 lines: jurisdiction, type of writing, client, opposing party, main claim, legal basis, deadline. Plus the list of indexed case-file exhibits.
2

Firm-style calibration + case-law lookup

Agent calls `search_style_house` to fetch 2-3 recent firm writings (citation conventions, tone, idioms). Then `search_jurisprudence` for relevant rulings to cite.
3

Structured draft generation

RAPPEL DES FAITS (each statement cited), DISCUSSION (legal syllogism), PRÉTENTIONS, DISPOSITIF - in CPC format. Each inference marked `[to validate]`, each exhibit cited by number.
4

Human arbitration + signature

The lawyer reviews the 20% needing judgement (tactical choice, tone, validation), corrects, signs. The agent always ends on `Maître [lawyer - TBC]` - no automatic signature.

Architecture

Tools, connectors, deployment.

Tools (function-calling)

search_pieces
search_jurisprudence
search_style_house

Optional connectors

sharepoint
outlook

Each connector activates based on the customer's subscription.

Deployment via the SDK

$ lmbox agent deploy ./conclusions-drafter \
    --box BOX-XXX \
    --token "$LMBOX_BOX_API_KEY" \
    --api https://api.lmbox.ai

LMbox guarantees across the catalogue

Data stays with you

Model and data stay on the customer's LMbox appliance. No patient, contract or invoice data is ever sent to an external cloud.

Audit chain

Every tool call, every agent output is timestamped, hashed and admissible before the regulator (ACPR, ANSM, CNIL, EBA).

Human decision

The agent recommends, the human decides. No auto-signature, no auto-payment: final responsibility stays with the business.

Other catalogue agents

Law firms pilot

NDA Reviewer

Reads each incoming NDA, identifies non-standard clauses against the firm's internal template library, drafts an amendment memo for the responsible partner.

1 h 30

Measured gain · par contrat

2 tools 2 connectors

v0.1.0 See the agent

All sectors pilot

Meeting Summarizer

Turns a raw meeting transcript (Teams, Zoom, locally-transcribed audio) into a structured CR: decisions made, action items with owner and deadline, open questions.

20-40 min

Measured gain · par réunion

3 connectors

v0.1.0 See the agent