Novelty — Design Mockup v1 VAULT DIRECTION · CLICK TABS TO WALK THE JOURNEY
Confidential prior-art screening

Is your idea actually new? Find out — without telling the world.

Drop in your sketches, specs, or diagrams. Novelty searches 150M+ patent documents worldwide and shows you the closest prior art, feature by feature, with evidence you can verify — all inside a sealed, zero-retention workspace.

Nothing stored after your session Never used to train AI No human ever sees your files
New invention check Sealed session
Drop your invention documents here
or browse files — sketches, specs, diagrams, photos of a napkin
PDFPNGJPGDOCXMDTXT
Files are encrypted in transit, processed in an isolated environment, and permanently deleted when you close your session — unless you choose to save your project.
AES-256 encryption at rest & in transit Zero-retention by default Invisible to everyone — including us Built with a registered patent lawyer
Why confidentiality comes first

An unprotected disclosure can kill a patent. So we built the vault before the search engine.

Telling the wrong tool about your invention can count as public disclosure — or hand your idea to a model that trains on it. Novelty is engineered so your submission is never stored, never seen, and never learned from.

Zero retention, by default

Your documents live in encrypted, session-scoped storage and are permanently destroyed when you leave. Saving a project is an explicit choice — and deleting it is one click, verified, with no backups quietly kept.

Sealed processing

Analysis runs in an isolated tenant — your material never mingles with anyone else's, never reaches advertising or analytics systems, and is never used to train any AI model. Ours or anyone's.

No human eyes

No employee, contractor, or reviewer can open your submission. If you later choose attorney review, you decide exactly what is shared, with whom, and under what engagement — nothing moves without your explicit approval.

"We designed Novelty the way a law firm treats a client file: sealed, privileged in spirit, and shredded when you say so."
Read the security architecture →
How it works

From napkin sketch to evidence-linked report in about twenty minutes.

1

Drop your documents

PDFs, diagrams, photos, specs — whatever you have. Novelty reads text and figures and asks a few sharp clarifying questions.

2

Approve your fingerprint

We distill your idea into an "invention fingerprint" — the essential features, in your words. You edit and approve it before anything is searched.

3

We search the world

Hybrid search across EPO, USPTO, and WIPO collections — 150M+ documents — grouped by patent family and ranked by real technical similarity.

4

Read the evidence

A feature-by-feature comparison against the closest references, with quoted passages and links — plus what still looks distinct about your idea.

A screening,
not a verdict.

What Novelty gives you

  • The closest prior art, with quoted evidence you can check yourself
  • A clear map of which features look known and which look distinct
  • An honest account of confidence levels and search limitations
  • Attorney-ready materials that cut hours off professional review

What no software can give you

  • A guarantee that your invention is patentable
  • A legal opinion — that requires a qualified patent professional
  • A complete search of every disclosure ever made, in any language
  • A substitute for filing advice on timing, scope, and jurisdiction
Pricing

One annual plan, billed monthly. Search as your ideas arrive.

Every plan is a full-year contract billed monthly, with the same sealed, zero-retention workspace at every tier.

For individual inventors
Individual Basic
$99/mo
10 searches per month · annual contract, billed monthly
  • Full multi-database prior-art screenings
  • Invention fingerprints with evidence-linked reports
  • Feature-by-feature comparison matrices
  • PDF export & encrypted project storage
Choose Basic
For teams, firms & portfolios
Enterprise Basic
$999/mo
10 seats · 100 searches per month
  • Shared team workspaces
  • Role-based access controls
  • Organization-wide audit logs
  • Centralized billing & admin
Choose Enterprise Basic
Enterprise Unlimited
$15,000/mo
Unlimited seats · unlimited searches
  • Everything in Enterprise Premium, without limits
  • Custom retention & deletion policies
  • Regional or private hosting options
  • Dedicated account team & security review support
Talk to sales

All plans are full-year contracts billed monthly. Attorney review and filing coordination are separate professional services performed by registered patent professionals — and if you ever file, the government fee, platform fee, and professional fee are always shown as separate line items, with no hidden markups.

Projects / Acoustic pipe-leak detector
1Upload 2Fingerprint 3Prior art 4Report

Add your invention documents

Drop in anything that describes the idea — polished or not. Novelty reads text, diagrams, and photos, then asks only the questions it can't answer from your files.

Continue to fingerprint →
Drop more files, or paste a description
Everything is encrypted the moment it leaves your machine
PDF
leak-detector-technical-spec.pdf
14 pages · 2.1 MB
Analyzed
PNG
sensor-placement-diagram.png
Diagram · 6 components, 8 relationships found
Analyzed
JPG
whiteboard-photo-jan14.jpg
Photo · handwriting detected
Reading…
MD
notes-on-ml-model.md
3.4 KB
Analyzed
Clarifying question · 2 of 3
Your spec mentions the model runs "on the device itself." Is on-device processing essential to the invention, or could the classification also happen in the cloud?
Answers sharpen the search. Skip anything you're unsure of.SkipAnswer

What's happening

~2 min remaining
  • Documents encrypted & sealedAES-256, isolated tenant
  • Text & diagrams extracted3 of 4 files complete
  • Identifying inventive conceptsProblem, mechanism, components
  • 4Building your invention fingerprintYou review before any search runs

Nothing leaves the vault

Extraction runs inside your sealed session. No search has been sent to any external database yet — that only happens after you approve the fingerprint in the next step.

Projects / Acoustic pipe-leak detector
Upload 2Fingerprint 3Prior art 4Report

Your invention fingerprint

This is how Novelty understood your idea — every field traced back to your documents. Correct anything that's off. The search only runs on what you approve.

One-sentence summaryHigh confidenceEdit
A battery-powered sensor that clamps onto residential water pipes and uses on-device machine learning to distinguish leak vibrations from normal flow — alerting homeowners before damage occurs, with no plumbing modification and no internet required.
Drawn from technical-spec.pdf, p.2 and notes-on-ml-model.md
Technical problemHigh confidenceEdit
Existing leak detectors either require cutting into the pipe (flow meters), detect water only after it has already pooled (moisture sensors), or stream raw audio to the cloud — raising cost, privacy concerns, and failure risk when connectivity drops.
Drawn from technical-spec.pdf, §1.2 "Background"
Core mechanismHigh confidenceEdit
A piezoelectric accelerometer captures pipe-wall vibration; a quantized neural network running on a low-power microcontroller classifies the vibration signature against a per-household baseline learned during a calibration period, discriminating leak turbulence from appliance use, water hammer, and thermal expansion.
Drawn from technical-spec.pdf, §3 and sensor-placement-diagram.png
Essential featuresEdit
Clamp-on mounting, no pipe modification On-device ML classification Per-household adaptive baseline Piezoelectric vibration sensing Offline operation + Add feature
Optional featuresEdit
Mesh networking between sensors Automatic shutoff-valve trigger Battery-life estimation + Add feature
Likely classificationsMedium confidenceEdit
G01M 3/24 — leak detection by vibration G06N 3/08 — neural network learning F17D 5/06 — pipeline monitoring

Before we search

  • Calibration period unclear. Is the baseline learned over hours, days, or continuously?
  • No dates provided. Have you shown this publicly, sold it, or posted about it? Timing affects your filing options.
  • Sensor coupling. Does the clamp require acoustic coupling gel, or dry contact?

Search on your approval only

Only the fingerprint above — never your original documents — is used to query patent databases. Queries are anonymized and carry no identifying information.

Approve & search prior art →
Projects / Acoustic pipe-leak detector
Upload Fingerprint 3Prior art 4Report

Closest prior art

Searched EPO, USPTO & WIPO collections · 41,208 candidates screened · 14 patent families ranked by technical similarity to your approved fingerprint. Similarity is an estimate — always read the evidence.

Generate report →
14
Relevant patent families found
2
References with substantial feature overlap
Novelty risk — review
3 of 5
Essential features not found together in any single reference
Potentially distinct
Search coverage
Patents & applications, EN/DE/FR, 1990–present. Non-patent literature not yet searched — included in the full report.
Ranked references All families Overlap ≥ moderate
Pipe-mounted vibration sensor with neural-network leak classification
US 10,845,270 B2 · Flowserve Corp. · Priority 2018-03-12 · FAMILY OF 4 · Active
High overlap4 of 5 essential features
Your essential featureIn this reference?Evidence
Clamp-on mounting, no pipe modification"…a housing removably secured to an exterior surface of the conduit by a resilient clamp…" — claim 1, col. 6
Piezoelectric vibration sensing"…a piezoelectric transducer generating a signal responsive to conduit wall vibration…" — claim 1
On-device ML classificationNeural network disclosed, but processing occurs "at a remote server" — col. 9, ln. 14–31. Edge processing not taught.
Per-household adaptive baselineUses a fixed factory-trained model; no site-specific calibration disclosed.
Offline operationRequires network connectivity for classification and alerting.
Disclosed Partially disclosed Not found
What may set you apart

This reference classifies in the cloud with a fixed model. Your combination of on-device inference with a per-household learned baseline operating fully offline does not appear in this family — though the combination must still be assessed for obviousness against multiple references together.

Acoustic monitoring system for water distribution networks
EP 3 421 950 A1 · Aquarius Spectrum · Priority 2017-06-28 · FAMILY OF 6 · Granted
High overlap3 of 5 essential features
Machine-learning water event detection from flow signatures
US 2021/0102863 A1 · Phyn LLC · Priority 2019-10-04 · FAMILY OF 3 · Pending
Moderate2 of 5 essential features
Self-calibrating vibration sensor for appliance diagnostics
WO 2020/174 512 A1 · Bosch GmbH · Priority 2019-02-25 · FAMILY OF 5 · National phase
Moderate2 of 5 essential features
Clamp-on ultrasonic flow measurement device
US 9,752,908 B2 · Siemens AG · Priority 2014-11-19 · FAMILY OF 2 · Active
Lower1 of 5 essential features
Projects / Acoustic pipe-leak detector
Upload Fingerprint Prior art 4Report

Prior-art screening report

Every assertion below links to the underlying patent passage. Download it, keep it, or hand it straight to a patent professional.

Download PDF Share with an attorney
Novelty.

Prior-Art Screening Report
Acoustic pipe-leak detector

Prepared 28 Jul 2026Databases: EPO · USPTO · WIPO41,208 candidates screenedRef. N-0042

Executive summary

Fourteen relevant patent families were identified. Two — US 10,845,270 B2 (Flowserve) and EP 3 421 950 A1 (Aquarius Spectrum) — disclose substantial portions of the invention, including clamp-on piezoelectric vibration sensing with neural-network classification. However, no single located reference discloses the combination of on-device inference, a per-household adaptive baseline, and fully offline operation. This combination is the strongest candidate for a distinguishing contribution, subject to an obviousness assessment by a qualified professional.

At a glance

2
References with substantial single-source overlap
14
Relevant patent families examined in detail
3
Essential features not found combined in any reference

Potentially distinguishing features

① Edge inference on a low-power microcontroller where the art teaches remote-server classification · ② a per-household baseline learned in situ, versus factory-fixed models · ③ classification and alerting with no network dependency. Each is supported by quoted passages in Section 4 of the full report.

Limitations of this screening

Search covered patent publications in English, German, and French from 1990 onward. Non-patent literature (papers, products, standards) was screened only at headline level. Unpublished applications less than 18 months old cannot be searched by anyone. Similarity ranking is probabilistic and may miss art describing the same idea in unusual vocabulary.

This report is not a legal opinion. It is a software-assisted screening intended to inform your next steps and reduce professional review time. Patentability determinations — including obviousness, eligibility, and claim scope — require assessment by a qualified patent professional.

Suggested next steps

  • 1Document your invention dates and any public disclosures — timing can limit filing options.
  • 2Generate an invention-disclosure package so an attorney starts from organized material, not a blank page.
  • 3Have a registered patent professional assess obviousness over Flowserve + Bosch in combination.

Turn this into an attorney-ready disclosure package

Structured disclosure, draft claim concepts with fallbacks, inventor questionnaire, and this report's evidence — exported to DOCX & PDF. Typically saves 3–5 hours of billed attorney time.

Included with Individual Premium & Enterprise plans
Build my disclosure package

Done with this project? Close it and everything — documents, fingerprint, results — is permanently destroyed. Or save it, encrypted, under your control.