Docs

Documentation

How Lexnus works, what the words mean, and how to connect a surface to the engine. Start with the loop below — everything else is reference.

Start here

Three steps from an empty account to your policy being enforced. No implementation project, no configuration call.

1

Build a playbook from a contract you already have

Upload one signed contract or one Word template. Lexnus reads it, extracts the clauses and the standards implicit in the language — the liability cap you accept, the governing law you prefer, the clauses you always include — and drafts a playbook.

A lawyer on your team reviews it, adjusts what is wrong, and publishes. Nothing is enforced until a human publishes it.

2

Analyse a counterparty contract against it

Upload inbound paper. The engine extracts the Smart Fields your rules reference, evaluates every rule, and returns a verdict per rule: pass or fail, with the contract language that caused it.

The result is written to the record, tied to the exact playbook version that produced it.

3

Connect a surface

Point another tool at the same engine. The application, the Word add-in, or an AI agent over MCP all ask the same question and get the same answer — because the playbook, not the surface, decides.

See Surfaces for what is live and what is on the roadmap.

Concepts

Seven words carry the whole model. Getting these right makes everything else obvious.

ClauseA unit of approved contract language, reviewed and published by a lawyer. Clauses live in the Clause Library. Nothing outside the library gets assembled into a contract.
RuleA plain-language condition your policy requires, written once. Aggregate liability must not exceed 1.5x annual contract value. Rules reference Smart Fields, so the comparison is precise rather than interpretive.
PlaybookA governed set of rules and clauses that defines one legal policy — your NDA standard, your MSA standard. A published playbook is immutable. Editing it creates a new version.
Smart FieldA structured data point extracted from contract text: a cap amount, a notice period, a governing law. The AI extracts the value. The rule evaluates it.
AnalysisOne contract evaluated against one playbook. Also the unit your plan meters.
VerdictThe result of one rule against one contract: pass or fail. Deterministic. There is no score, no confidence, no suggestion — the playbook decides, not the model.
The recordThe append-only history of what was asked, what was answered, and which frozen policy version answered it. This is what makes a decision defensible a year later.

Surfaces

A surface is anywhere a contract question can be asked. Every surface calls the same engine, so the answer does not change with the tool.

Lexnus applicationBuild playbooks, run analyses, browse the record. The full interface.
Word add-inAnalyse and draft inside the document, where lawyers already work.
MCP-connected agentsClaude, ChatGPT, Mistral and any other MCP client. 23 tools. Reference.
Read-only APIGrowthQuery playbooks, clauses and analysis results. Read and write on Enterprise.
Outlook add-inRoadmapCheck attached paper without leaving the mail client.
Public write APIRoadmapFull read and write access outside Enterprise agreements.

API and MCP reference

The MCP reference is the complete technical document: all 23 tools with JSON Schema, the authentication model, scopes, rate limits and the audit behaviour of every write. The essentials:

Server URLhttps://app.lexnus.com/mcp
TransportStreamable HTTP. Stateless — every call is authenticated independently.
AuthenticationOAuth 2.1 authorisation code flow, or a personal access token issued in Lexnus
Scopeslexnous:read, lexnous:write
Tools23 — twelve read, eleven write
Tool manifestGET https://app.lexnus.com/mcp/tools
Rate limit5 requests per second sustained per account, burst 20
AnalysisAsynchronous. Trigger, then poll for the result.

MCP integration overview covers the same ground for a non-engineering reader.

Feature guides

Legal PlaybooksHow rules are written, versioned and published.
Contract AnalysisWhat happens to an inbound contract, rule by rule.
Clause LibraryGoverning the language itself.
Contract DraftingAssembling outbound paper from approved clauses only.
MCP IntegrationConnecting AI tools to your policy.
AI speed without AI riskWhy the model reads and the playbook decides.

Something missing here?

If you needed an answer this page did not give you, tell us. Gaps in the docs are bugs.