Straight answers about Patent Bear.
Where the data comes from, how the app is hosted, what AI sees, what is private, and how the patent workflow fits together.
Data and coverage
Where does Patent Bear get its patent data?
Patent Bear gets its raw data from public patent-office records, including USPTO sources, but the important part is what we do next: we host the patent data ourselves. We build our own searchable corpus of patents, publications, applications, file histories, assignments, family data, PTAB records, figures, OCR, citations, and generated research artifacts.
That is why Patent Bear feels different from a patent portal or a chatbot wrapper. The data is not somewhere else. It is inside the product, connected to the reader, search engine, citations, annotations, claim charts, Ursa, and Research analyses.
Because we host the corpus, we can make patent records interactive, searchable, citable, and usable by AI workflows in ways that tools built on links and pasted text cannot.
What patent records can I search?
Patent Bear supports searching US patents, pre-grant publications, patent applications, and related technical sources. The difference is that search is not bolted on as a separate feature.
The same hosted corpus powers number lookup, advanced search, cited Patent Chat, Ursa search planning, annotations, saved work, and Research analyses. When you find a record, you can immediately read it, cite it, chart it, ask questions about it, inspect its file history, and run analysis over it.
Are patents and applications always free to look up?
Basic patent, publication, and application lookup is designed to be free. Some saved workspace features, AI chat, claim charts, team features, and Research analyses may require an account or paid plan because they use storage, compute, and AI budget.
Do you host official patent PDFs?
Official patent PDFs remain patent-office records, and PDF downloads may link to or come from the original source where available. Patent Bear hosts the layer professionals actually work with: structured text, claims, figures, OCR, column-line artifacts, search indexes, citations, and generated research files.
That hosted layer is what turns a static PDF into a workspace. You can jump from an AI citation to the source, keep figures beside claim language, annotate passages, build claim charts, and run analyses against the same record.
How current is the data?
Patent Bear is designed to stay current with public patent-source data because the product depends on having a living hosted corpus, not a stale static dump. Source systems can still lag or change, and official records should be checked for legal deadlines, ownership, and status. But our architecture is built around continuously turning public patent records into searchable, usable workspace data.
Does Patent Bear replace USPTO, Patent Center, Global Dossier, or PTAB systems?
No. Official portals remain the legal source of truth for filings, deadlines, and status. Patent Bear is different because it is built for doing the work after you find the record.
Official systems store documents. Patent Bear hosts and structures the data so you can search across it, read it cleanly, annotate it, cite it, ask AI about it, generate claim charts, and run analyses without rebuilding context in a dozen tabs.
AI, Ursa, and Research
Do you use my data to train AI models?
No. Patent Bear does not use your prompts, chats, annotations, saved patents, analysis inputs, analysis outputs, invention descriptions, product descriptions, or work product to train AI models.
When AI providers process API requests for Patent Bear, their current API terms state that API inputs and outputs are not used to train models unless the customer opts in. Patent Bear does not opt in.
Beyond that, every AI call Patent Bear makes runs under Zero Data Retention: OpenAI and DeepInfra do not log or keep your prompts or results after answering the request.
What gets sent to AI providers?
For AI chat and analyses, we may send the patent text, selected claims, pasted invention or product descriptions, generated elements, search queries, evidence snippets, and your instructions as needed to provide the requested analysis, both to the model that writes the response and to the reranker that ranks which patent passages are most relevant to your question. Billing details, passwords, and account credentials are not sent to our AI providers.
Is Ursa a generic chatbot?
No. Ursa is what becomes possible when AI is connected to a hosted patent corpus instead of an empty prompt box. A generic chatbot only has whatever you paste into it.
Ursa can work from patent text, claims, figures, prosecution history, family data, citations, search results, and analysis run context already inside Patent Bear. The advantage is not just a better model; it is a better working environment for the model.
Are AI answers legal advice?
No. Ursa, Patent Chat, and the analyses in Research are research tools. They can miss references, misunderstand claim language, cite the wrong location, or produce incomplete analysis. Verify every citation, date, claim mapping, legal standard, and conclusion against the underlying documents and qualified counsel before relying on it.
What is Research?
Research is where the hosted corpus becomes a real workflow advantage. Prior art, invention search, disclosure support, clearance screening, claim construction, and priority analysis all need more than a prompt.
They need search, ranking, claims, specifications, citations, file history, generated elements, and a place to save the result. Patent Bear combines those pieces into structured workflows that produce reviewable outputs instead of one-off chatbot answers.
Why do analyses cost credits?
An analysis can run many searches, ranking passes, OCR or column-line jobs, and AI calls in a single workflow. Credits cover that compute and AI usage while giving you a predictable budget.
Privacy and security
Where is Patent Bear hosted?
The Patent Bear web application is hosted on Vercel. Account data and application data live in Supabase. Patent figure images and related generated files are stored in Cloudflare R2 and served through Cloudflare.
Some worker services for analyses, OCR, column-line generation, search, ranking, and analysis run on Patent Bear's own private, single-tenant Hetzner-hosted infrastructure, not a shared multi-tenant pool.
Where is my account data stored?
Account data, saved patents, annotations, highlights, notes, search history, chats, analysis runs, and durable outputs are stored in Patent Bear's database and object storage systems. Worker services are designed as transient compute, not durable storage for user work product.
Is my data encrypted?
Yes. Patent Bear uses HTTPS for data in transit. User data is stored with encrypted infrastructure. The Security page describes provider controls, encryption, authentication, worker processing, and access-control practices in more detail.
Do you sell my data or use advertising trackers?
No. Patent Bear does not sell your data to advertisers or data brokers. We do not use advertising pixels or cross-site tracking scripts.
Privacy-friendly analytics, when enabled, is limited to aggregate product usage and excludes raw patent queries, pasted work product, credentials, and billing details.
Who can see my saved patents, annotations, chats, and analysis runs?
Your saved patents, highlights, notes, chats, search history, and analysis runs are private to your account unless a team-sharing feature is enabled and you choose to use it. Administrative access is limited to operational need, debugging, reliability, billing, abuse prevention, and security response.
Can I export or delete my data?
Yes. You can request export or deletion of account data. Account deletion removes active application records, subject to limited backup retention and operational requirements described in the Privacy Policy.
Workflow
What can I do without an account?
You can search and read public patent records without an account. An account lets Patent Bear remember your work: saved patents, annotations, highlights, notes, claim charts, chats, analysis runs, usage, and plan state.
What does the Library do?
The Library is the research hub for your account. It brings together recent activity, saved patents, annotations, search history, claim charts, chats, analysis runs, plan state, and usage so you can resume work without reconstructing it from browser history.
Can I annotate patents?
Yes. You can highlight text, save claim terms, add notes, annotate figures, track occurrences, hide or restore markings, and keep research state attached to a patent. Signed-in users can persist annotations across sessions.
Can I export claim charts?
Yes. Patent Bear can split claims into elements, let you add notes beside each element, and export charts to Word for briefs, memos, or client reports. Claim chart limits may depend on your plan.
Does Patent Bear support teams?
Patent Bear is designed for patent professionals and IP teams. Team collaboration features can support shared annotations and workspace access depending on plan and setup.
How do I get help?
Start with the Help page for feature documentation. If something is wrong or unclear, contact Patent Bear with the patent number, application number, page, or analysis run so the issue can be reproduced quickly.
Need a deeper answer?
The Help, Security, Privacy, and Terms pages have the longer-form version of these answers.