Biotech patent intelligence

We read every page. Check any value we deliver.

WO 2021/041671

KRAS G12D inhibitors · Table 2 · p. 829

Human-reviewed
CompoundIC₅₀ (nM)

Source · two pages joined

Compound 1 · 124.7 nM

Binding to KRas G12D by exemplary compounds of Formula (I)

DeliveryExport to CSV or Excel today, with SDF and API workflows next.Reported in nM · units normalised
WO 2021/041671 · check every figure in the public analysis
Extraction tables
0
Compounds
0
Values with source evidence
0

Jubust turns the patent PDFs you select into structured compounds and reported IC50 values. Every datapoint in the delivered dataset is human-reviewed and linked to its source.

See example data

Inspect the dataset. Open the source behind any value.

This public KRAS example links each compound structure and activity value to the exact patent page and evidence region. Select a linked value to open the PDF source drawer.

Public KRAS dataset
Swipe table → · Tap a value for PDF source
Structure
KRAS G12D
KRAS G12D
KRAS G12D
KRAS G12D
KRAS G12D
Table-only preview from WO2023018699Go to demo app

Missing data is not just inefficient. It's a risk.

Hidden data in hundreds of pages

Relevant activity is buried in tables, figures, examples, and attachments.

Inconsistent units and assay conditions

Different units, formats, and assays make comparison risky.

Broken context from cross-references

Important details are spread across claims, examples, and figures.

Missing values lead to wrong conclusions

One missed value can invalidate priorities and decisions.

Manual reconstruction creates review work

Re-keying tables and following cross-references creates opportunities for omissions.

Patent search stops too early.

Most tools help teams find documents. Jubust turns patent PDFs into verified, source-linked SAR datasets your scientists can trust and analyze.

Extract compounds and bioactivity
Generic
Partial
AI-first
Partial
Jubust
Source table labels and assay endpoints
Generic
Partial
AI-first
Partial
Jubust
Source traceability
Generic
×
AI-first
Partial
Jubust
Human verification
Generic
×
AI-first
×
Jubust
Structured exports (CSV, XLSX)
Generic
AI-first
Jubust
Find relevant patents
Generic
AI-first
Jubust
Roadmap
Cross-patent compound index
Generic
×
AI-first
Partial
Jubust
Roadmap

Extracted by pipeline. Verified by humans.

Our pipeline extracts compounds, assay values, and source locations from dense patent PDFs. Reviewers then verify the SAR record and can see the exact PDF location behind every value.

Patent WO2023018699

Pipeline extraction pass

6 spans linked
Source snippetHuman-verified SAR row
01FROMTABLE1SourceWO2023018699SOURCE
02SOURCEPATENTCompoundEXAMPLE1ASSAYROW
03TABLE1KRAS WTKRASWTIC50ASSAYVALUEIN
04RESULTGREATERValue10000NMOBSERVED
05KRAS G12DKRASG12DIC50INASSAYDATARANGE
06RESULTValue45NMCONFIRMEDASSAYLOGS
07SOURCEPagePAGE125PATENTREFERENCE

Human-verified SAR row

Compound + activities + source

Structure

Example 1 compound structure

Compound ID

Example 1

Activities

KRAS (WT) IC50 (nM)

>10,000 nM

KRAS (G12D) IC50 (nM)

45 nM

Source page

p. 125

View full page

What changes for your team

A reviewed SAR dataset, delivered.

Move selected patent PDFs out of manual spreadsheet reconstruction and into a structured, human-reviewed dataset.

Proof: the public KRAS record connects a compound structure and reviewed activity values to their patent pages.

Every value is click-verifiable.

Open the exact patent page behind each delivered compound structure and activity value.

Proof: the public KRAS example links its 8 nM result to page 126.

Per-project delivery.

Fit a bounded pilot into your existing analysis workflow without migrating platforms or buying seat licenses.

Proof: reviewed CSV/XLSX joins the tools your team already uses.

What you actually get

Structured SAR dataset

Structured rows with CSV and Excel export, ready for filtering, review, and analysis.

Source-linked evidence

Extracted values link back to the source page and evidence region in the patent.

Normalized bioactivity

Numeric assay values ranked and normalized to nM where units are parseable.

Expanding

Compound structures

Structure images and linked compound records, with identifier coverage expanding across filings.

Today

Assay labels and endpoints

Included today from source tables. Cell-line and method context remain on the roadmap.

Roadmap

Claim relevance

Claim mapping is planned for teams that need IP relevance alongside SAR evidence.

From selected patents to a reviewed SAR dataset.

1. Upload

Upload 10-50 patent PDFs around your target, disease area, or modality.

2. We extract and verify

Our pipeline extracts compounds and bioactivity. Reviewers verify every datapoint included in the delivered dataset.

3. You get your dataset

Receive a clean, structured, source-linked dataset ready for analysis.

4. Export and integrate

Export to CSV or Excel today, with SDF and API workflows next.

Runs exclusively on EU infrastructure. Your queries, internal compound IDs, and target lists never leave the region.
Built for pharmaceutical and chemical patent workflows, not generic document parsing.

Test Jubust on a patent slice your team already understands.Inspect the reviewed dataset against the source.

Request a sample dataset

Bounded scope. Human-reviewed. Source-linked.