Ground a legal assistant
Give your chatbot real jurisdictional context and cited sources instead of best guesses. Every answer traces back to the graph.
retrieve · cite
Enterprise plan only
Caseworth Context API
Legal intelligence, as an API your AI can call.
Ground your own assistants, agents, and products in Caseworth's legal knowledge graph. One request returns jurisdiction-aware context, benchmark ranges, and cited sources. Legal information, not advice.
Provisioned per organization · SOC 2 Type II · Your data is never used for training
14
States Live
2.4M+
Indexed Matters
<200ms
p50 Retrieval
14 states
Live, rolling out nationally
2.4M+
Matters in the graph
50+
Matter types classified
Daily
Source refresh cadence
What you can build
Build legal-grade products, without building the graph.
The Context API is one endpoint over the same legal knowledge graph that powers Caseworth. Call it from your assistants, agents, and internal tools.
Estimate case value
Pull benchmark ranges for matters like the one in front of you, framed the careful way: cases like yours have ranged from X to Y.
benchmarks
Triage & classify intake
Turn free-text descriptions into a matter type, jurisdiction, and routing signal your product can act on in real time.
classify
Check jurisdictional fit
Resolve which courts and rules apply before you route, file, or quote. Built state by state, returned as structured data.
jurisdiction
How it works
A retrieval layer, purpose-built for law.
One request in, grounded context out. The graph does the reasoning about jurisdiction and matter relationships so your team doesn't have to.
01
Send a query
Post a plain-language description or structured fields. No schema gymnastics. Short input is fine.
02
We traverse the graph
Caseworth resolves jurisdiction, matter type, and the relationships between them across the legal knowledge graph.
03
You get grounded context
Structured JSON: classification, benchmark ranges, jurisdictional fit, and citations you can surface or feed to a model.
04
Cite, don’t hallucinate
Every fact carries a source reference, so your product stays defensible and your users stay informed.
query · traverse · ground · return
The API
One endpoint, in your stack today.
Authenticate with an org-scoped key, post a query, and get grounded context back. SDKs for the languages your team already ships in.
curl https://api.caseworth.com/v1/context/retrieve \ -H "Authorization: Bearer $CASEWORTH_KEY" \ -H "Content-Type: application/json" \ -d '{ "query": "rear-end collision, herniated disc", "jurisdiction": "FL", "include": ["benchmarks", "citations"] }'
{ "matter_type": "personal_injury / motor_vehicle", "jurisdiction_fit": "FL · 11th Circuit", "benchmark": "cases like yours have ranged from X to Y", "citations": [ { "source": "statute", "ref": "FL §627.737" }, { "source": "matter_set", "ref": "fl-mva-2k" } ], "confidence": 0.89 }
What's in the graph
Structured legal context, not a pile of documents.
The Context API returns clean, typed fields your model can reason over and your engineers can rely on.
matter_type
Matter classification
Free text resolved to a taxonomy of 50+ matter types and sub-types, from employment to personal injury.
jurisdiction
Jurisdiction resolution
State, district, and venue mapping for every state we’ve rolled out, returned as structured fields.
benchmark
Benchmark ranges
Outcome ranges framed carefully: cases like yours have ranged from X to Y. Never a single promised number.
citations
Cited sources
Every response carries source references so your product can show its work and stay defensible.
- matter_type
- employment · wrongful_termination
- jurisdiction
- TX · Southern District
- benchmark
- ranged from X to Y
- citations
- [ 14 sources ]
- confidence
- 0.91
- updated
- 2026-06-09
Legal information, not advice. Responses are context for your product to use. They don't create an attorney-client relationship.
Why Caseworth
The bridge from a question, to the right answer.
Teams choose the Context API because the hard part, keeping a national, jurisdiction-aware legal graph fresh and cited, is already done.
Jurisdiction-aware by design
Built state by state as a national platform. Context is scoped to where your matter actually lives.
Fresh, not frozen
Sources refresh daily. You’re grounding on current law and current ranges, not a stale snapshot.
Cited and defensible
Every fact traces to a source. Your product can show provenance and your team can stand behind it.
One graph, two surfaces
The same knowledge graph that bridges people to firms now backs your tooling. One source of truth.
Careful by construction
Benchmark framing and UPL-safe phrasing are baked into responses, so you inherit the guardrails.
Built for builders
Predictable JSON, low-latency retrieval, and SDKs for the languages your team already ships in.
Enterprise
Access for teams who answer to regulators.
The Context API is available on the enterprise plan only. That comes with the controls a legal-grade integration demands.
SOC 2 Type II
Independently audited controls across security, availability, and confidentiality.
Isolated by org
Keys, usage, and context are scoped per organization. No shared tenancy of your requests.
Never trained on your data
Your queries and payloads are never used to train models, ours or yours.
Audit logs & SLAs
Full request audit trails, role-based access, and contractual uptime for enterprise teams.
Get access
Available on the enterprise plan.
The Context API is provisioned per organization with usage-based pricing, a coverage scope for the states you operate in, and a dedicated integration contact. Tell us what you're building and we'll get you keys.
Questions
What teams ask before they build.
Product and AI teams building legal-grade tools: assistants, intake systems, case-evaluation products, and internal agents. It's the same legal knowledge graph that powers Caseworth, exposed as one API.
Yes. The Context API is available on the enterprise plan only. That's deliberate. It comes with the data isolation, audit logging, and SLAs a legal integration requires.
Caseworth is a national platform rolling out state by state. Coverage is scoped to the states you operate in at provisioning time, and grows as we roll out. We'll confirm your coverage during onboarding.
Usage-based, on top of the enterprise plan, with volume tiers. Pricing depends on coverage scope and request volume, and your integration contact will scope it with you.
No. Your queries and payloads are never used to train models, ours or anyone else's. Context is scoped per organization and isolated from other tenants.
No. Responses are legal information, not advice, and don't create an attorney-client relationship. Benchmark and UPL-safe framing is built into the payload so your product inherits the guardrails.