API & MCP
Industry leading patent search, in your stack.
Integrate the same models that power Clerq into your products, platforms, or agents. Semantic search and analysis across patents and non-patent literature, over a REST API and an MCP server.
"Even if I had the best prompt in the world, I'll never get good results if the initial population is wrong. You have to have a good provider. We just went through a round of evaluations, tested Clerq against the others, and it came out on top."
why Clerq api & mcp
Every workflow that needs patent intelligence faces the same problems — Maintaining a massive patent database and indexing pipeline, Machine Learning expertise needed to train patent specific models, and building a secure API layer to wrap it all up. The Clerq API & MCP solves them all, for commercial products and internal tools alike.
Skip the hard part
The embeddings, training, and tuning — the years we've obsessed over getting this right, ready to deploy.
Managed for you
The corpus stays current and model keeps improving — no pipeline to babysit, no tech debt to inherit.
Built for volume
Consistent latency across millions of calls, sized for agentic traffic.
How it Works
The model embeds your query, searches patents and non-patent literature by concept, and returns a ranked list of prior art.
Send a query over REST or MCP
The NLPatent model searches by concept
Get structured, citable JSON back
The saved list builds itself as you work — export to Word, Excel, or straight into an IDS.
REST API
Patent search in a few API calls.
A clean HTTP API over the model — send natural language queries or patent numbers, get back ranked prior art your code can use. Each result arrives with its metadata attached.
Sandbox first
Credentials, documentation, and a sandbox to prototype before you commit.
Structured JSON
The same shape in every response — parse it once, build on it.
Search-as-a-Service, not data
Ranked results only — bulk data, full text, PDFs and figures require a data provider.
MCP SERVER
Patent tools for your own agents.
The MCP server hands your agents patent search as a native tool — any MCP-compatible model can call it, and a general-purpose agent gains a specialist's retrieval.
Any MCP client
Plug into Claude, Cursor, or your own agent.
Corpus-grounded
Tool calls return ranked results that trace back to the references they came from.
Composable tools
Callable functions your agent can chain into bigger jobs.
THE MODEL
The engine behind Clerq, without the UI.
You build on the exact model Clerq’s own agents use — encoder-only, trained only on patents, in production with top IP teams. Enterprise security and data handling included.
Patent-native
Encoder-only and patent-trained — retrieval can't hallucinate a document.
In production
Serving top law firms, IP teams, and software vendors reliably, for years.
Enterprise-ready
ISO 27001 and SOC 2 Type II, with data isolation and dedicated support.
“Clerq has gone from helping me search to doing the work under my direction. What comes back minutes later reads like an expert did it.”
500M
Patent and non-patent literature documents from every major jurisdiction searched before Clerq answers
5 Years
Building a model that understands patents the way you do — the foundation our agents stand on
2000+
Professionals at top law firms and innovative companies have made Clerq part of their practice
Built on Clerq's API & MCP
From customer-facing features to custom internal workflows, the same calls do the work — whatever sits on top.
Automated prior art search
Trigger a search straight from a disclosure or claim set, inside your own review flow.
Platform integration
Patent retrieval inside legal tech platforms, specialist or general — users never leave.
RAG and Agent pipelines
Ranked, traceable patent context for grounding what your models generate.
Custom internal tools
Disclosure intake, FTO screening, and competitive tooling on your own stack.
Book a demo and see Clerq in action.
Book a demo and see Clerq in action.