Search first, build second
Months of R&D can already be sitting in one of 51M+ real patents. See who has done what, how far they got, and where the whitespace is — mine innovation directions with TRIZ, grow your idea into an R&D roadmap, and design around blocking patents.
Four R&D tools that turn ideas into solid research
Use millions of real patents as ammunition to help R&D teams find directions worth pursuing and make them solid — we don't write patents for you, we help you do the research.
Got a working solution but unsure where to break through next? Use TRIZ to systematically find R&D directions worth pursuing.
- 1Describe your technical solution, or the contradiction you're stuck on
- 2AI breaks down the contradiction, applies 40 inventive principles, and searches real patents for breakthroughs
- 3Get a list of R&D directions, ranked by value
Directions aren't guesses — they're the white space found after seeing the existing art clearly.
Want to know what others have already done in a direction? Semantically search tens of millions of real patents in one line, see the prior art, then decide where to go.
- 1Describe the technology in plain language, or look up a patent number directly
- 2Semantic search across three real patent libraries: CN inventions (21.8M), CN utility models (22.8M), US (6.4M)
- 3Get a relevance-ranked list of real patents, and generate a landscape report in one click
Every result is a real, existing patent — not AI-generated; the report only synthesizes what was retrieved.
One idea, millions of real patents as ammunition — developed into an actionable R&D roadmap.
- 1Describe your idea, or paste your disclosure (the more real and specific, the better)
- 2AI searches real patents, maps the landscape, and plans the implementation path
- 3Get a 5-part R&D roadmap, with unverified items clearly left blank
This is an R&D roadmap, not a patent application; anything you haven't done is marked 'verify-in-R&D' — never fabricated.
Blocked by a competitor's patent? Break its claims down element by element, find the weakest limitation, and take a different technical route to a better product — designing around patents is legitimate R&D, not a shortcut.
- 1Free scan: enter competitor patent numbers, or describe the competitor and the product you want to build
- 2AI maps every claim element, locates the weakest limitation, and generates design directions each self-checked under the doctrine of equivalents
- 3Get a product design reference ready for your R&D backlog: design options, feasibility, trade-offs, and a pre-launch verification checklist
This is a product design reference and R&D record, not a legal opinion: risks are graded as ranking signals only — we never declare non-infringement; unverified parameters are marked 'verify in R&D', never fabricated. Engage counsel for a formal FTO before launch.
See a sample report first →Research direction set, ready to file? Estimate official and service fees for China, US, and Europe online — the firm handles the full process as required, with a US-registered patent practitioner on the US side.
Full-chain representation from patent mining and drafting to filing and office-action responses (unlimited rounds); transparent official fees, service fee is the firm's formal quote — we act as required and make no promise about examination outcomes.
Use a real patent database to do real R&D
Not another 'AI writes your patent' tool — we help R&D teams do real research with millions of real patents.
Tell us your R&D problem
Describe your real idea or challenge in your own words — even if it's still rough.
Search & compare against real patents
AI searches 51M+ real patents, finds the gaps, and locates white space — our moat: a real, searchable patent vector database.
Get structured R&D output
Innovation directions or an R&D roadmap, with unverified items clearly left blank — never fabricated.
“A few blocking competitor patents, one different technical route — a better product, and a seven-figure sum (RMB) of duplicated R&D avoided.”
— A manufacturing business owner, after a counsel-led design-around engagement
Individual results vary by project · The outcome came from a counsel-led professional service; this platform is a productized aid built on that methodology, and its output still requires patent counsel's review
Your data is protected: documents are only used to process your request and are not used for AI model training
Explore R&D Directions
Already have a working solution? We'll help you systematically explore R&D directions you might have missed.
Patent Knowledge
Frequently Asked Questions
Everything you need to know about Smart Patent
What is Smart Patent and who is it for?
Smart Patent is an AI-powered patent-intelligence R&D assistant designed for R&D engineers, tech leads, founders, and innovation teams. Grounded in a real, searchable patent database, it helps you mine for innovation and turn ideas into an R&D roadmap — so you get your R&D direction clear before engaging a patent attorney.
Can Smart Patent replace a patent attorney?
No. Smart Patent is an R&D preparation and analysis tool that helps you clarify your R&D direction before working with a patent attorney. It automates prior art search, feasibility analysis, and R&D direction planning, but final patent filing should always involve a qualified patent attorney.
What types of patent analysis does Smart Patent support?
Smart Patent currently offers four core public features: Innovation Mining (TRIZ-based innovation directions), Patent Search (semantic search over tens of millions of real patents), R&D Exploration (turning an idea into an R&D roadmap), and Design-Around Studio (breaking down competitor claims to generate a product design reference that routes around infringement). All are built on a real, searchable patent database, helping R&D teams do real research and product design with patent intelligence.
How accurate is the AI analysis?
Smart Patent uses multiple leading AI models optimized for the patent domain. While AI analysis provides valuable insights and catches issues human reviewers might miss, results should be treated as professional-grade references rather than definitive legal opinions.
Which patent offices are supported?
Smart Patent's patent search covers massive patent data from CNIPA (China National Intellectual Property Administration) as well as major patent offices worldwide (USPTO, EPO, JPO, KIPO, etc.). Both Innovation Mining and R&D Exploration build on this real, searchable patent library, with deep optimization for Chinese-language patent documents. The platform supports both Chinese and English interfaces.
What file formats are supported?
Smart Patent's analysis features accept PDF and Word (DOCX) file uploads, and you can also simply paste or type a text description.
What's the difference between Innovation Mining and R&D Exploration? Which should I use?
Innovation Mining (TRIZ) is for when you already have a solution and want to find the next breakthrough — it starts from technical contradictions and gives you multiple innovation directions. R&D Exploration is for when you have an idea and want to develop it into a complete, actionable R&D roadmap. Both are grounded in real patent search.
Is your patent search real? Will it make things up?
Search is grounded in a real, verifiable patent database. Any parameters, materials, or values the AI cannot derive from your input are clearly marked 'verify-in-R&D' and never fabricated; patent numbers not found in search results are redacted. This respects R&D rigor and is also a compliance baseline.
How is this different from just asking ChatGPT or a general AI?
General AI often fabricates patent numbers. We retrieve from a real database of 51M+ patents (CN inventions, utility models, US), so every result is an existing patent you can verify. It's 'can vs cannot', not just 'better'.
How large is your patent database?
About 51M+ patents: ~21.8M CN inventions, ~22.8M CN utility models, and ~6.4M US patents, with search coverage spanning CNIPA, USPTO, EPO, JPO and KIPO.
How does TRIZ-based Innovation Mining work?
You describe your solution or the contradiction you're stuck on. It analyzes the technical contradiction and proposes multiple differentiated R&D directions, each checked against existing patents to find white space.
What do I get from R&D Exploration?
One idea becomes an actionable 5-part R&D roadmap, grounded in real patent search. Anything the AI can't derive is left blank as 'verify-in-R&D' — never fabricated.
Do I need patent expertise to use it?
No. Describe what you're building, testing, or improving in plain language — even a rough idea. The tools handle the patent search and analysis for you.
Is the technical content I enter kept secure?
Your inputs are used only to run your analysis. We don't sell your data, and unverifiable values are never invented into your output. See our Privacy Policy for details.
Is it free? Do new users get credits?
New users get 25 free credits to try the features. After that it's pay-as-you-go credits with no subscription, and credits never expire.
Is it for individuals and small teams, or only large companies?
It's built for R&D engineers, tech leads, founders, and innovation teams of any size — from a solo inventor to an enterprise R&D department.
Is design-around legal?
Yes — it is competition that both US and Chinese patent systems explicitly encourage: the public-notice function of claims exists precisely so competitors know where the boundary is and can innovate outside it. Designing around is R&D in nature: a different technical route to an equal or better product. Note: our output is an engineering reference; a formal FTO by patent counsel is still required before launch.
Do you guarantee a design-around won't infringe?
No — and no tool should make that guarantee. What we do: every candidate direction is adversarially attacked by our own doctrine-of-equivalents engine, and directions that fail are explicitly flagged; surviving candidates only get ranking signals to help you decide which direction to take to counsel first. Honest signals beat false guarantees.