Legal · 12 weeks to production

Cutting first-pass contract review time by 82% with cited AI

An AI contract review assistant that flags risky clauses across thousands of agreements, with every flag citing the exact clause text, so lawyers verify in one click instead of re-reading.

faster first-pass contract review: 82%; faster first-pass contract review
contracts processed monthly: 3,000+; contracts processed monthly
of flags backed by clause citations: 100%; of flags backed by clause citations

The challenge

Client: Mid-size corporate law firm
Services: AI & Machine Learning, Custom Software
Timeline: 12 weeks to production

Associates were spending 60% of their week on first-pass review: reading NDAs and vendor agreements clause by clause to find the handful that deviate from firm playbooks. Expensive hours, repetitive work, and the error rate climbed every Friday afternoon.

The partners had rejected two AI tools already, both for the same reason: answers without evidence. A flag a lawyer can't verify instantly is slower than no flag at all, because now they have to re-read everything anyway.

What we did

The approach, decision by decision

01
Built the playbook into evaluation sets first
Turned the firm's negotiation playbooks into 600 graded test cases before writing the product, so every model and prompt change was measured against what the partners actually consider risky.
02
Made citations the contract, not a feature
Every flag links to the exact clause span it came from. If the model can't ground a finding in quoted text, the finding doesn't ship to the lawyer.
03
Designed for review speed, not chat
No chatbot. A review screen that orders flags by playbook severity, shows the deviation side by side with the standard position, and exports straight to the markup document.

The results

First-pass review time down 82% across NDAs and standard vendor agreements.
3,000+ contracts a month processed, with flag precision above 95% on the eval suite.
Every flag citation-backed; partner spot checks found zero unverifiable findings.
Associates redeployed to negotiation and client work, the billable hours that matter.

Built with

Claude API
Python
FastAPI
pgvector
PostgreSQL
AWS

Keep reading

More case studies

Fintech · Consumer lending fintech

Cutting loan decisions from 4 days to 7 minutes

A digital lending platform with automated underwriting that took loan decisions from a 4-day manual review to a 7-minute automated flow.

Read case study

Logistics · Freight operations startup

Scaling a logistics SaaS from zero to 50,000 shipments a month

A multi-tenant freight management platform launched in 12 weeks, now processing 50,000+ shipments a month across three markets.

Read case study

Your project could be the next case study

Tell us the metric you need to move. We'll scope the build that moves it: fixed price, weekly demos, results you can publish.

Book my free consultation Send a project inquiry

We reply within 1 business day.