Contract review agents that run inside your walls
Bespoke redlining agents that run inside your infrastructure. Grounded in your contracts and playbook, integrated in your existing tool set and CLM.
Four phases, three months, one flat fee
Corpus and playbook intake
We index your executed agreements, fallback positions and prior redlines out of your DMS. Nothing leaves your tenancy.
Agent build
Clause classifiers, deviation detection and drafting policies written against your standards plus public market clauses.
Deploy behind your controls
Runs in your VPC or on-prem, on your identity provider, your retention rules and your audit logging.
Evals and upkeep
A held-out set of your own matters scores every release. We maintain and re-tune the evals as your positions change.
A tracked-changes redline, with the reason attached
Every edit is a Word tracked change, citing the clause in your playbook or the executed precedent it came from. Reviewers accept, reject, or ask for the fallback.
Every pass is version-controlled, so you can move back and forth to the best version of the contract, with a clean audit trail of who changed what, and why, for the file.
Your servers, your protocols, your audit trail
In-tenancy runtime
Deployed to your VPC, private cloud or on-prem hardware. No document text transits our infrastructure.
Your model choice
Bring your Azure OpenAI, Bedrock or self-hosted weights. We build against the endpoint your security team already approved.
SSO, RBAC, retention
Your identity provider, your matter-level permissions, your deletion schedule, with full action-level logging.
Integrations
Practice management and CLM: Clio, iManage, NetDocuments, SharePoint, Ironclad, Agiloft, and a Word add-in.
I'm a data engineer with 15+ years building the systems that hold, move, secure and delete personal data, and the last 6+ years spent applying that work inside legal tech. I've led data engineering at Apple, Intuit, Dropbox, and Clio.
Before Clio, I was the founding data engineer at Lawyaw, a document-automation platform for lawyers that Clio acquired. At Clio, I helped build and scale Vincent AI, which grew revenue from $200M to $500M ARR in under nine months.
Flat fee, agreed before we start
No seat licences and no per-document metering. One implementation fee, then hourly upkeep for as long as you want the evals maintained.
- ·Corpus indexing and playbook encoding
- ·Agent build, integration and in-tenancy deployment
- ·Eval suite on your own matters, handed over to you
Maintaining the systems and continuously updating the evals as your clause positions, counterparties and tooling change.