Case Watch
Persistent monitoring of federal matters for new docket activity.
Re-checked about every 6 hours; alerts by email.The litigation data layer for AI
The infrastructure connecting litigation to every AI agent.
CaseMagic turns fragmented court records, case events, correspondence and litigation relationships into persistent, structured intelligence that lawyers, businesses and AI agents can use.
Federal litigation is the wedge. The Litigation Knowledge Graph is the platform.
- 94Federal district courts
- Every U.S. district court, territorial courts included, searchable by case number.
- 88Court filing feeds read
- Districts whose own new-filing feeds CaseMagic reads every five minutes.
- 16,523New federal cases, 30 days
- New civil cases read from the courts’ feeds and indexed in the last 30 days.
- 125,727Public case pages
- Case pages published for search engines and AI crawlers.
- 17MCP tools
- Litigation tools any MCP-capable assistant can call at casemagic.ai/mcp.
- 3Paid plans, billing live
- $99, $299, $599 per month through Stripe.
Counted from production. Revenue and customer figures are shared in the investor deck.
The problem
Litigation data was built for humans browsing databases. AI needs infrastructure.
The record of a single lawsuit is scattered across systems that were never designed to talk to each other, and none of them was designed to be read by an AI agent.
- Court systems
- PACER
- RECAP and public archives
- PDF filings
- Correspondence
- Calendars
- Case-management products
- Private notes
- Several AI assistants
CaseMagic
One persistent, structured record per matter, with identity and context an agent can call.
AI agents can reason about litigation only when litigation has persistent structure, identity and context.
Architecture
CaseMagic is the litigation data layer for AI.
Courts create the authoritative litigation record. Lawyers and litigants create the private matter record. AI agents need structured, persistent access to both. CaseMagic connects them.
Sources
- Federal court feeds
- Dockets and filings
- RECAP archive
- Case correspondence
- Uploaded documents
The layer
CaseMagic- Ingestion
- Normalization
- Case Passport
- Deadlines
- Entity resolution
- Monitoring
- Intelligence
- Permissions
- Litigation Knowledge Graph
Consumers
- ChatGPT
- Claude
- Gemini
- Grok
- Law firms
- Companies
- Legal software
Solid: in production. Dashed: being built with this round.
The loop
Every monitored case makes the network more useful.
Court events happen whether or not anyone opens the software. That makes litigation a recurring data stream, and a monitored case a recurring reason to come back.
- Public court data
- Normalization
- Case Graph
- Monitored matters
- Private correspondence
- AI interactions
- Structured intelligence
- Better discovery
- More users
- More monitored matters
↺ and back to step 01, with more context than the last pass.
Court documents can be replicated. A continuously enriched litigation graph that combines public records, private matter context and agent interactions is designed to become progressively harder to reproduce. That is the defensibility we are building, not one we claim to have finished.
Product
Five products on one data model.
Each one is live at casemagic.ai today. The interface fragments below are drawn from the product; the case shown is illustrative.
Case Passport
Persistent, structured identity and context for each matter.
One record per matter that survives a change of hands or assistant.CaseMail
Private correspondence connected directly to the matter.
A permanent address per case; messages and attachments filed to it.Filing Streams
Continuous discovery of newly filed federal litigation from reusable criteria.
Read from 88 district courts’ own filing feeds.Agent / API layer
Structured litigation access for AI assistants and applications.
17 MCP tools, listed in the official MCP registry.The product in eight steps
From a case number to an agent that knows the case.
Every step below runs in production. Check a real case free; it takes under a minute.
- Search a caseAny federal district case, by number. No PACER login.
- Open the Case PassportCaption, judge, status and the latest filings in one record.
- Turn on Watch7-day trial, then a monthly plan.
- A new filing arrivesDetected on the next pass and sent as an alert.
- The record updatesThe docket, deadlines and correspondence stay together.
- Ask your assistantChatGPT, Claude, Gemini or Grok, through MCP.
- Fresh context comes backThe assistant reads CaseMagic, not a pasted paragraph.
- ActDraft the response, calendar the deadline, tell the client.
Distribution
Every major AI assistant can become a CaseMagic distribution channel.
AI assistants are not competitors to CaseMagic. They are where people already ask about their cases, and each one that can call CaseMagic is a channel that brings users to the record.
- ChatGPTLive
CaseMagic app in ChatGPT; real lookups from ChatGPT’s own client recorded in production.
- ClaudeVerified over MCP
Remote MCP with OAuth, PKCE and dynamic client registration, run end to end against production.
- GeminiLive via function calling
Gemini called CaseMagic’s check_case function and answered from live docket data.
- GrokVerified over MCP
The free lookup runs over the stateless MCP transport xAI documents.
- MuseNot yet tested
Same MCP and OAuth shape as Claude; not yet connected from Muse itself.
User, in an AI assistant
“What changed in Acme Corp. v. Smith?”
The assistant calls CaseMagic
get_case_updates · the docket, new filings, deadlines, parties, correspondence and the Case Passport
The assistant answers
With structured, current litigation context, and a link back to the case in CaseMagic.
Then
The user discovers CaseMagic through the agent. Every link an assistant prints carries a signed source, so a customer who arrives through ChatGPT is credited to ChatGPT in our attribution, not to “direct”.
Statuses come from the compatibility record behind the integrations page: “Live” means real use from that assistant’s own client has been recorded; “Verified over MCP” means the protocol path has been run end to end against production.
Traction
A working system, shipped in weeks.
Revenue, customer and retention figures are in the investor deck. What is public is the system itself, and how fast it was built.
Shipped
- First code, and live federal case lookups the same day.
- Live billing: three plans at $99, $299 and $599 a month with a 7-day trial.
- CaseMagic app submitted to OpenAI for the ChatGPT directory.
- Gemini calls CaseMagic and answers from live docket data.
- “Is this company being sued?” answered from the courts’ own feeds, as an MCP tool.
Why now
AI is changing the interface. Litigation infrastructure has not caught up.
Agents act on our behalf
People increasingly expect an assistant to reach into specialized systems for them, not to be pasted into.
Legal AI is a funded category
Institutional capital has validated legal AI as a major software category. The data layer under it is still open.
MCP made it callable
The Model Context Protocol lets ChatGPT, Claude and other assistants invoke external structured services with the user’s permission.
The record is still fragmented
The authoritative litigation record remains structurally hard for a general-purpose AI to monitor persistently.
Market
Wedge, then platform, then infrastructure.
Built bottom-up from who pays for each layer, not from one speculative TAM figure.
Wedge
U.S. federal litigation monitoring
Self-represented litigants and small practices with active federal matters, paying monthly per monitored matter.
Professional
About 1.3 million U.S. attorneys
Plus paralegals and litigation support, per the ABA’s national lawyer population count, on the higher-priced plans.
Enterprise
Institutions with litigation exposure
Law firms, corporations, insurers, financial institutions, compliance teams, litigation finance and legal software vendors.
Infrastructure
Applications and agents
Software and AI agents consuming litigation intelligence through the API and MCP, priced on usage.
Expansion beyond the wedge: state courts, then administrative and regulatory proceedings, on the same data model.
Business model
Recurring revenue that follows the case.
A monitored case is a subscription that lasts as long as the litigation. Expansion comes from more matters per account and from higher tiers.
Individual
$99per month, live today
One monitored matter.
Professional
$299per month, live today
Up to 10 monitored matters.
Practice
$599per month, live today
Up to 50 monitored matters.
Enterprise / Infrastructure
Contractusage-based, planned
For firms, companies and applications calling the API at volume.
Category
A different category, not a cheaper tool.
Legal AI assistants reason over what they are given. Practice-management software organizes a firm’s workflow. Court databases provide documents and dockets. CaseMagic maintains the persistent, structured litigation context that people, applications and AI agents can keep reading.
| Capability | CaseMagic | Docket monitoring | Practice management | Legal AI assistant | Public court database |
|---|---|---|---|---|---|
| Persistent monitoring | Yes | Yes | No | No | No |
| Private correspondence context | Yes | No | Yes | Partly | No |
| Persistent matter identity | Yes | Partly | Yes | No | Partly |
| Litigation knowledge graph | Building | No | No | No | No |
| AI-agent access | Yes | No | Partly | Yes | No |
| API / MCP | Yes | Partly | Partly | No | Partly |
| New-filing discovery | Yes | Partly | No | No | Partly |
| Cross-case intelligence | Building | No | No | No | Partly |
| Public + private context together | Yes | No | No | No | No |
| Actionable alerts | Yes | Yes | Partly | No | No |
Categories, not named products. “Building” marks what this round funds.
Defensibility
From application to infrastructure.
Each layer depends on usage of the one before it, so the stack compounds as monitored cases, correspondence and agent calls accumulate.
- Litigation Knowledge Graph — the structured relationships between matters, parties, counsel, judges and filings
- Normalized historical events — every docket event in one shape, across districts
- Persistent Case Passports — the identity of each matter over its whole life
- Private CaseMail context — correspondence no public database holds
- Agent interaction data — which questions agents ask about which cases
- Derived litigation intelligence — deadlines, explanations, clusters of related suits
- External-agent integrations — ChatGPT, Claude, Gemini, Grok and what follows
- Search and AI citation authority — public case, company and court pages that answer engines cite
- Workflow persistence — a monitored case stays monitored until the case ends
- Distribution — search, agents and product-led acquisition together
Expansion
One data model, more jurisdictions.
Each step adds courts to the same model of cases, parties, events and documents. It is not a new product per jurisdiction.
- NowFederal district courts
All districts, live.
- NextFederal appellate and bankruptcy
Same national case-number system, same feeds.
- Phase IIIMajor state courts
California, New York, Texas, Florida and Delaware.
- Phase IVNationwide state litigation
Led by real unsupported-case demand from users.
- Phase VAdministrative and regulatory proceedings
The same graph, beyond the courtroom.
Capital efficiency
Small team. Infrastructure-scale ambition.
Capital is being used to accelerate a working system, not to discover whether a product can be built. Live court lookups, billing and the MCP endpoint were all in production within the first week.
AI-assisted engineering
A small team ships at the pace of a much larger one.
Product-led acquisition
The free case check is the top of the funnel.
Machine and agent distribution
Public pages for search, MCP for assistants.
Automated onboarding
Sign up, check a case and start a trial with no call.
Self-service subscriptions
Stripe checkout, no sales cycle below enterprise.
Infrastructure reuse
Every court added feeds every product at once.
No large outbound sales organization is planned below enterprise contracts.
The round
Raising $3M Seed to make CaseMagic the default litigation layer for AI.
$3,000,000
- 30% $900,000Product & Engineering
The core platform, the Case Graph, enterprise controls and the intelligence layer.
- 20% $600,000Data Infrastructure
Court ingestion, normalization and jurisdiction coverage.
- 20% $600,000Growth & Distribution
Durable search, agent and product-led acquisition channels.
- 10% $300,000AI-Agent Ecosystem
Deep integrations across AI assistants and agent ecosystems.
- 7.5% $225,000Security & Compliance
Enterprise trust, controls and compliance readiness.
- 7.5% $225,000Sales & Customer Success
Selective high-value customer acquisition and expansion.
- 5% $150,000Legal, Admin & Reserve
Corporate operations and contingency.
What this round unlocks
- Coverage
Substantially expanded U.S. litigation coverage, starting with appellate and bankruptcy.
- Scale of the graph
Millions of additional normalized events and the relationships between them.
- Enterprise
Enterprise-grade deployment, controls and compliance readiness.
- Agent distribution
Deep, maintained integrations across the major assistants.
- Infrastructure revenue
API and MCP usage priced for applications, not just people.
- Measured acquisition
Agent-originated customer acquisition tracked end to end.
- Higher ARPU
More professional and practice accounts with more matters each.
- Institutional use cases
Insurers, litigation finance and compliance teams.
CaseMagic at a glance
Investment summary.
- Category
- AI-native litigation infrastructure
- Business model
- Recurring SaaS, with enterprise and API layers
- Current wedge
- U.S. federal litigation
- Platform asset
- The Case Graph, growing into the Litigation Knowledge Graph
- Distribution
- Search, AI agents and product-led growth
- Round
- $3M Seed
- Capital objective
- Expand the litigation graph, distribution and jurisdiction coverage
Investor FAQ
Questions investors ask.
What exactly is CaseMagic?
CaseMagic is the litigation data layer for AI. It monitors federal court cases, captures case correspondence, keeps one persistent structured record per matter (the Case Passport), and makes that record available to people and to AI assistants such as ChatGPT and Claude through an MCP endpoint.
Why isn’t this just another legal AI assistant?
A legal AI assistant reasons over what it is handed. CaseMagic is what it is handed: the persistent, continuously updated record of a case that any assistant can call. Assistants are distribution channels for CaseMagic, not competitors.
Where does the litigation data come from?
Federal dockets and filings come from CourtListener and its RECAP archive, and new cases come from the district courts’ own public filing feeds, read every five minutes. Private context comes from the user: correspondence sent to a case’s CaseMail address and documents they add.
What is the CaseMagic Litigation Knowledge Graph?
It is the structured graph CaseMagic is building of matters, parties, counsel, judges, filings, deadlines, correspondence, companies and related cases. Today the Case Graph links courts, cases, docket events, documents, deadlines, correspondence and the companies and plaintiffs named across new federal suits; this round funds extending it into the full Litigation Knowledge Graph.
What is proprietary?
Not the public filings, which anyone can obtain. What is proprietary is the structure built on top of them: normalized events, persistent Case Passports, the private correspondence attached to each matter, the record of how agents use it, and the relationships derived across cases.
Does CaseMagic integrate with ChatGPT?
Yes. CaseMagic is available to ChatGPT as an app over MCP, and production has recorded real case lookups from ChatGPT’s own client. Claude, Grok and Muse connect the same way, and Gemini through function calling.
Why does CaseMagic need AI-agent integrations?
Because that is where people now ask about their cases. Every assistant that can call CaseMagic is a channel that brings users to the record, and every link an assistant prints is attributed back to it.
How does CaseMagic make money?
Monthly subscriptions per monitored matter: Individual $99, Professional $299 and Practice $599 per month, with a 7-day trial. Enterprise contracts and usage-based API pricing are the next layers.
How much revenue does CaseMagic have?
CaseMagic shares revenue and customer figures with investors directly, in the investor deck.
What is the initial market?
Active U.S. federal civil litigants: self-represented litigants, small litigation practices and small firms with active federal matters.
Why start with federal litigation?
One national case-number system across every district, public filing feeds and one consistent docket format. That makes it the fastest place to prove the data model before adding state courts.
What prevents an incumbent from copying this?
Nothing stops anyone copying the software. What is hard to copy is a graph enriched continuously by monitored cases, private correspondence and agent usage, together with distribution already established inside the assistants. That position compounds with usage, which is why speed matters now.
How is customer data protected?
Every case belongs to one organization and every request is answered only for the organization that owns it. An assistant reaches a private matter only after the user signs in to CaseMagic and authorizes it over OAuth. Pages behind a sign-in carry a no-index header, and payments are handled by Stripe. The security page sets out what is stored, who can reach it and what CaseMagic does not claim.
What will the $3M Seed fund?
Product & Engineering 30%, Data Infrastructure 20%, Growth & Distribution 20%, AI-Agent Ecosystem 10%, Security & Compliance 7.5%, Sales & Customer Success 7.5%, Legal, Admin & Reserve 5%.
How can an investor request the deck?
Use “Request Investor Deck” on this page. Only a name and an email address are required; the founder sends the deck personally.
Litigation will become agent-native.
CaseMagic is building the infrastructure layer that makes that possible.
Courts create litigation. CaseMagic structures it. AI agents consume it.