Mid-market FinServ edition
Drop counterparty paper in. Get a track-changes redline back in 90 seconds. Send it for signature without leaving Word. Built for legal-ops teams who'd rather close deals than chase committee approvals.
SOC 2 Type 1 in flight · On-prem available · AWS, Azure, GCP
Three things you can't get anywhere else
Ironclad, Icertis, Spotdraft, and Evisort are great at the contracts you draft. They struggle on the larger half — the contracts counterparty sends you. That's where we live.
Our LLM agent reads counterparty paper, applies your playbook clause by clause, and returns a track-changes DOCX redline plus a drafted negotiation email — in 90 seconds. Every proposed edit cites the playbook rule that drove it. Auditable. Defensible. Forensic-grade.
Lawyers live in Microsoft Word. Our add-in mounts inside Word
with native Office.js — proposed edits land as real
<w:ins> / <w:del> revisions
in the Review ribbon. Accept, reject, or modify each one. The
workflow your partners already know.
Production tenant ready 48 hours after contract signature. Your backlog of 10,000 legacy contracts auto-classified overnight with 90%+ accuracy. Your first playbook drafted from your own executed contracts. Most CLM implementations take six months because they have to; ours doesn't.
How it works
Drop a PDF or DOCX. Our extraction pipeline pulls 50+ fields per contract type — parties, term, governing law, indemnity caps, payment terms — with confidence scores. Anything below 85% gets routed to human review.
Your playbook lives in the platform — ideal positions, fallback positions, walkaway points, severity per rule. The agent reads counterparty's draft, identifies every deviation, proposes a tracked-changes edit for each one with rule-cited rationale.
Proposed edits land in Microsoft Word as native track-changes revisions. Your senior partner steps through Accept / Reject just like any other Word redline session. Twenty seconds of review on three edits, versus an hour of manual drafting.
DocuSign and Adobe Sign integrated natively. Send the redlined version with one click — envelope status streams back to the platform via signed webhooks. No tab-switching.
Same AI extracts obligations: quarterly reporting, SLA commitments, audit-right windows. Renewal alerts fire at 90, 60, and 30 days before auto-renew. Slack, email, Teams. Nothing falls through.
Screenshot placeholder
Drop assets/screenshots/redline-preview.png here.
See README.md.
Platform breadth
Full lifecycle coverage from intake to obligations. Same engine, same audit trail, same workflow runtime. The wedge is what makes the platform earn its keep at $300K ACV.
LLM agent applies your playbook, cites rules, drafts negotiation emails.
Native Office.js, tracked-changes revisions, AppSource-ready.
50+ fields per contract type. JSON-schema, function calling, confidence-gated.
BM25 + dense vectors with cross-encoder reranking. Citations on every answer.
Six pre-built templates. Edit visually in Camunda Modeler. Flowable engine.
Tenant-defined per-contract-type fields. Text, number, date, select, multi-select.
Reporting due dates, SLAs, audit-right windows — all extracted, assigned, tracked.
90/60/30-day windows. Slack, email, Teams. Nothing auto-renews without you knowing.
DocuSign and Adobe Sign integrated. Status streams back via signed webhooks.
Okta, Azure AD, Ping. SCIM provisioning round-trips. Self-hosted Keycloak option.
Every state change. Each row's hash includes the previous — tamper-evident.
Cycle time, AI spend per tenant, redline conversion. Prometheus + Grafana, open standards.
Security & compliance
Per-tenant encryption with envelope keys. Hash-chained audit log partitioned monthly and archived with Object Lock. Defense in depth — tenant isolation at controller, repository, and object-storage prefix layers. SOC 2 Type 1 in flight, Type 2 observation period started.
Air-gapped on-prem deployment supported. Helm chart + signed image bundle. vLLM with Llama 3.3 70B for tenants who can't send contracts to Anthropic or OpenAI. Same code, same workflow, your VPC.
Pricing
We never charge per reviewer. Adding lawyers to the workflow is how you get value out of CLM; we won't fine you for it.
$75–180K/year
$500M–$1.5B revenue · 50–150 contracts/month
Most common
$300–600K/year
$1.5B–$5B revenue · 150–500 contracts/month
Custom
Regulated · air-gapped · banks & insurance carriers
Frequently asked
Ironclad's AI extracts data and surfaces clause-library matches — useful, but table stakes. It doesn't redline counterparty paper end-to-end with rule citations and a drafted negotiation email. We do. Spotdraft, Icertis, and Evisort are in the same camp. Harvey is closer in capability but doesn't run the CLM around it.
Every edit has a confidence score from the model. Anything below 85% gets human review by default — configurable per tenant. Every edit also has a rationale citing the playbook rule that drove it, so the reviewer can spot bad applications quickly. The AI proposes; the lawyer signs off. Same workflow as a junior associate's first-pass redline.
Cloud SaaS deploys in AWS US-East-1 or US-West-2 (your choice). On-prem deploys in your VPC, fully isolated. We do not train on your contracts. Default LLM is Claude via Anthropic's API; if your CISO requires no cloud LLM, we ship vLLM with Llama 3.3 70B running on-prem in your tenancy.
Production tenant ready within 48 hours of contract signature. Includes your tenant schema provisioned, your first playbook drafted from your existing executed contracts (using our auto-build tool), your SSO integrated, your Salesforce or Workday connector tested. Two weeks to team-wide comfort. Compared to Ironclad's six-month standard implementation.
No. Your data is never used to train upstream models. Per-tenant golden eval sets are stored locally to validate playbook performance — those stay inside your tenant and are not shared cross-tenant unless you explicitly opt in.
Salesforce ships in Phase 1. Microsoft 365 (SharePoint, OneDrive, Outlook), Google Workspace, DocuSign, and Slack are also Phase 1. Workday and SAP Ariba are Phase 3 (months 7–9). For the long tail of integrations we ship a Workato / Tray.io / Zapier connector by month 9.
Migration is part of onboarding. We auto-classify your legacy contract corpus during ingestion — 90%+ accuracy out of the box, confidence-scored, with the lower-confidence ones queued for human review. Most migrations finish in under 30 days.
Forty-five minutes, your screen-share, your counterparty paper. Watch the agent work in real time against your playbook. If you're not convinced inside the first ten minutes, we'll cut the call short.
We respond within one business day. No marketing spam — your email isn't going on a drip list.