One engine for novelty, prosecution,
and IPR-grade prior art.

An AI search engine that finds prior art across patents, journals, and clinical trials in one query, at the disclosure desk, in prosecution, or at the litigation file.

The same context search runs from inventor disclosure through prosecution to IPR. At the disclosure desk it tells you whether a draft is worth filing in 90 seconds. In prosecution it surfaces the office-action references your associate-led search missed. In IPR it builds a defensible invalidity record. One engine, three moments, scoped to the level of rigour each one needs.

Picture this

Two moments, one search. *Disclosure desk:* the inventor has spent six months on this, the room is celebratory, no-one has checked the prior art yet. *Prosecution:* the examiner cites two references you'd never seen, one in Japanese, one a 2017 conference paper. The question 'did your prior art search miss these?' has the same answer: only if your search was Boolean.

Before Kwintely vs With Kwintely

Before
With Kwintely
Two separate searches: novelty pre-draft, prior-art post-draft
One engine, two output modes, disclosure desk and prosecution
Boolean keyword search across multiple databases
Context search across patents, papers, and clinical trials in one query
Misses translations, terminology variants, adjacent disciplines
Finds conceptual equivalents regardless of language or framing
Skim 200+ documents to find the relevant 5
Top results surfaced by relevance score, with passage indicated
Risk: undiscovered prior art surfaces at office action or IPR
Defensible, auditable search trail with date stamps and similarity scores

Teams already using Kwintely for Prior Art & Novelty Search

From description to result in three steps

Input your claim or disclosure

Paste the independent claim, a technical abstract, or an invention disclosure. No Boolean operators needed; no formal claim language required at the disclosure stage.

Context search across all three document types

Our AI searches 295M+ patents, scientific papers, and clinical trials simultaneously ... cross-language, cross-discipline.

Receive ranked prior art + novelty grid

Pre-drafting: novelty score + anticipating art per element. Prosecution / IPR: ranked citations with passages and similarity scores.

Four artifacts, one search

PDF

Prosecution-ready PDF

Top citations with publication dates, similarity scores, and the relevant passages quoted verbatim. For office-action responses, IPR petitions, or invalidity briefs.

HANDOUT

Disclosure-desk handout

1-page novelty score + file / re-scope / abandon recommendation, plus per-element novelty grid. For the inventor disclosure-review meeting.

BIB

BibTeX export

For academic citation work and patent-citation cross-referencing.

TRAIL

Date-stamped search trail

Defensible audit log, when the search was run, against which corpus, with which query.

What teams ask before they sign.

When do I run this at the disclosure desk vs at prosecution?

Same engine, two scopes. At the **disclosure desk** (pre-draft): inventor pastes the disclosure, you get a 90-second novelty score plus anticipating art per element, file / re-scope / abandon. Filters out 30–50% of disclosures before they cost €5–15K to draft. At **prosecution / IPR** (post-grant or in litigation): paste the claim, you get ranked anticipating art with date-stamps and passages quoted verbatim, defensible for office action, opposition, or inter-partes review.

Is the output ready for counsel review?

Every result includes the source document, publication date, similarity score, and the relevant passage quoted verbatim. The output is structured the way an associate-led search would be: ranked, cited, exportable. The AI is a mapping engine; the documents it surfaces are real. Counsel still makes the legal call; Kwintely does the search and citation prep.

How does this fit with my existing patent firm?

Most teams run Kwintely before engaging counsel, to scope the work and narrow the candidate set, then walk into the review with the search prep already done. Counsel spends their hours on judgement and the legal opinion, not on assembling the candidate list.

What about pricing and lock-in?

Three ways to work with Kwintely: the self-serve SaaS platform, scoped research-service projects where our team runs the analysis for you, or the API and MCP data layer for your own tools. Full output runs on a paid plan, the free tier is for trying basic searches, not complete deliverables. See the pricing page for current plans and terms.

Run the experiment, not the procurement.

Run it on the next inventor disclosure, or the next claim being challenged.

Counsel still signs off

Kwintely produces the structured search; your firm keeps ownership of the legal opinion. Nothing about your existing relationship has to change.

Run it on a real case

Pick the technology you'd already pay counsel to clear. Compare what Kwintely returns against what your firm would charge for. Decide on the second case, not the first.

No lock-in

Free tier covers the first basic experiment. Full output on paid plans. Fixed monthly fee on the way up. The cost of the test is one search.

See Prior Art & Novelty Search in action.

Run it on the next inventor disclosure, or the next claim being challenged.