Chemistry-heavy patent PDFs, turned into reviewable scientific evidence.
Jubust extracts exemplified compounds, assay results, and example-level disclosures from patent filings — each finding anchored to its exact source page. Every answer is inspectable.
From hard patent artifacts to clean high quality machine-readable data
The bottleneck is not finding patent documents. It is reconstructing compounds, assay results, formulation details, and cross-page context quickly enough for real scientific decisions.
KRas G12D KD
PDF p.2
Structure proof
PDF p.13
Table 1 · Row 1
Compound and assay records
Extract exemplified compounds, reported assay values, and example-level details into structured records tied to the original filing.
p.2
Structure diagram
p.13
Table 1 · KD
p.47
Assay conditions
Example 1 · 97.7 nM · KRas G12D
3 sources linkedPage-level provenance
Preserve provenance across structures, tables, footnotes, and formulation details instead of splitting the evidence into isolated extraction fragments.
Quality signals
1 flagged fieldCompound
Target
Value
Signal
High-quality scientific output
Improve result quality through active learning on the fields scientists challenge, correct, and rely on most.
What scientists use Jubust for
Questions that come up in real patent review.
Across competitor sets, patent families, and compound series, scientific teams need clear answers on what was made, tested, and shown. Jubust answers those questions with source-linked records scientists can inspect directly.
What did they actually make and test — not just claim?
Which compounds were actually exemplified, which assays were run, what results were reported, and where is that shown in the source? Jubust returns source-linked compound, assay, and example records as structured, machine-readable data with provenance that a scientist can review directly.
Current scope: extracting reviewable evidence from the patent text itself.
Source passage · p.13
Example 1. To a 50 mL flask was added intermediate 3 (2.3 g, 8.5 mmol) in anhydrous DMF (20 mL) under N₂. The mixture was stirred at 60 °C for 18 h. After workup and silica gel purification, KD = 97.7 nM as determined by TR-FRET binding assay for KRas G12D. Fractions were pooled and concentrated under reduced pressure to afford the title compound as a white solid (1.1 g, 47% yield).
Extracted record
VerifiedKRas G12D · KD
What evidence across relevant patents should change our decision on this target?
Across the patents relevant to this target, mechanism, or lead series, which findings should affect a competitive or experimental decision? Jubust assembles a review pack with cited records across the selected filings, not just a summary paragraph.
Supported by cross-patent evidence linking.
3 compounds shown · 12 total for this target
Who is patenting compounds close to our lead series?
Which patents disclose chemistry structurally close to the series your team is working on — and what exactly did they show? This becomes available once extracted structures and scaffold linking are consistently good at scale.
Query series
KRas G12D scaffold
Close disclosures
WO2023041892
p.5
WO2022098763
p.12
Requires structural extraction at scale — available as evidence coverage grows.
Reviewable, not summarised
Detailed provenance. Fast validation.
The workspace links compound structures, binding values, and assay context directly to the patent pages they came from. Click any citation badge to open the source. No black-box answers.
Patent Queue
4 filings
WO2021041671
12 compounds
WO2023189012
Parsing…
WO2022051443
8 compounds
WO2024071830
Queued
Binding affinity · KRas G12D KD
IC₅₀ 114 nM · selectivity confirmed
Structure
Record fields
Compound
Example 1
Target
KRas G12D
Assay type
TR-FRET
Patent
WO2021041671
Filing date
Feb 2021
Source pages
p.2 + p.13
PDF p.2
Structure · anchor region
PDF p.13
Table 1 · Row 1 · KD
WO2021041671 · p.2 of 1,247
Chemical structures section
Also cited
p.13 · Table 1 · KD column
Row 1 → 97.7 nM confirmed
What the evidence enables
More evidence unlocks harder questions.
These use cases become credible once evidence is structured and linked at scale — the natural next layer above reviewed evidence packs.
What compound families are heating up around this target?
Which scaffold or compound families are seeing rising patent activity around this target or modality? Each trend row backed by the source patents underneath — not naked counts.
3 new filings in the past 90 days · Each row links to source patents
Which programs are starting to show cell, in vivo, or translational evidence — not just potency tables?
Which patent programs are moving beyond binding or potency claims into stronger biological or translational evidence? Evidence-type tracking, not quality assessment.
Evidence type tracking only — does not imply scientific quality or program success.
Where this leads
A patent-native chemistry intelligence layer.
These are the questions worth asking once enough reviewable evidence is structured and linked. We are not there yet — but this is the direction that makes the current work worthwhile.
Where are the underexplored regions worth reviewing next?
Which parts of this chemical neighborhood appear thinly covered by current patent literature — and which nearby patents are worth deeper investigation? Framed as exploration and hypothesis generation, not definitive white-space discovery.
Chemical space · KRas G12D
Coverage reflects structured evidence only. Not a definitive prior-art landscape.
What evidence links this scaffold, target, assay, and assignee across the field?
Navigate connected scientific evidence — how is this chemistry connected across patents, targets, assays, and companies? Every edge resolves back to source evidence. Not abstract network art.
Assays
- TR-FRET
- Biochemical displacement
- Cellular viability
Assignees
- Amgen
- Mirati
- Rev. Medicines
Patent families
- WO2021041671
- WO2023189012
- WO2022051443
- WO2020186101
- WO2021252339
- WO2022147188
- WO2023034499
Each entity resolves to patent source evidence — not inferred relationships.
Common questions
Evaluate Jubust on the patents your team actually reviews.
Bring the compounds, assay evidence, and competitive questions that matter to your team. We scope design-partner engagements around reviewable output, not generic demos.