Capability  ·  01

Read every contract. Cite every flag.

The foundation underneath the rest of the platform. Five specialised agents parse, classify, and extract — into a structured model that the rest of Evaluetor can reason over.

01 / What it does

Six things, at portfolio scale.

Parties, dates, values

Names, signatories, jurisdiction, effective and termination dates, contract value, currency, payment terms — extracted from the document, not from a sidecar metadata form.

36 clause types

Indemnity, liability, IP, termination, confidentiality, assignment, force majeure, governing law, dispute resolution, change of control, exclusivity, audit rights, and 24 more — each classified, scored, and cited.

15 contract schemas

1,235 typed fields across MSA, SOW, NDA, employment, lease, SaaS, services, supply, distribution, licensing, M&A, JV, franchise, partnership, insurance.

Paragraph-level citation

Every extracted fact and every risk flag points back to the exact paragraph in the source document. Legal counsel can verify a finding in seconds.

Cross-portfolio anomaly

Find the one outlier indemnity clause across 200 SaaS contracts. Surface the supplier whose termination terms drift from your portfolio norm.

Contract Q&A

Ask natural-language questions across the portfolio. Answers cite source. Built on the same vector index used by extraction.

02 / Under the hood

Five agents in concert.

01
Metadata extraction

Parties, dates, values, jurisdiction

02
Schema extraction

15 contract types · 1,235 typed fields

03
Clause extraction

36 clause types · paragraph-citing

04
Contract Q&A

RAG-powered answers across the portfolio

05
Intent routing

Queries routed to the right agent or tool

Each agent's reasoning chain is written to a Langfuse trace — prompts, model output, citations — and can be replayed, exported, or audited.

Try it

Drop in five contracts. See what it pulls.