AI Drug Discovery Platform for Small Molecules

L Lead AI Discovery workspaceIllustrative workspace

SCREENING RUN / DEMO-026

A shortlist, with the evidence attached.

Ready for review

06 Example candidates04 Ranked for review02 Held outside domain

Candidate / sourceRelative score + uncertainty

Interval shown with every scoreIllustrative relative scale: 0–1

Select a candidate to inspect its reasoning.Sample records and scores · not experimental results

Supported workflowVirtual Screening for Small Molecules

Why teams come
to us for this.

Chemical space is too big to brute-force, and rankings built on contradictory assay data are hard to trust.

A ranked shortlist for lab prioritisation
A data card: origin, preparation, test split and version
Published: 21.6% lower model error and R² 0.87 on an MDM2 benchmark

FROM DATA TO A DEFENSIBLE DECISION

In silico drug discovery,
with the reasoning attached.

A useful prediction starts before the model.
And stays connected to its evidence after it.

01

Purpose-built data preparation

Structures and units standardised, duplicates resolved, provenance recorded in a data card.

Compare the incoming and prepared records

ASSAY RECORDS / EXAMPLE

Different files. One usable record.

CompoundMeasurementSource

CMP-01442 nMAssay A

CMP-01442 nMImport B

Units aligned. Duplicate linked to its original source.

DATA CARD / v1.4Preparation stays reproducible.

Origin · structure standardisation · units · duplicate history

Recorded

02

Selective cleaning

Conflicting measurements resolved by weight of evidence, not averaging.

Select a measurement to inspect its context

ONE COMPOUND / THREE RECORDSCMP-014

Illustrative measurements · evidence context drives the example decision.

ASSAY RECORD C / EVIDENCE REVIEW

A disagreement deserves a closer look.

The assay conditions differ from the comparison set. This value is retained in the source history and flagged for review, rather than silently averaged in.

Flag for review · preserve the source

Original value and decision both retained

03

Library prioritisation

Large collections narrowed to a ranked shortlist for the next decision.

Switch between the shortlist and review queue

Your candidate libraryCurated + repurposing collections

01Prepared data

02Model + uncertainty

03Applicability check

REVIEW QUEUE / EXAMPLE

01CMP-014Evidence + uncertainty attachedRanked

02CMP-028Evidence + uncertainty attachedRanked

03CMP-063Evidence + uncertainty attachedRanked

04CMP-091Evidence + uncertainty attachedRanked

Prioritised for a scientific decision, then lab work.

04

Reasoning and provenance

Every shortlist ships with the reasoning and provenance for review.

Explore the reasoning, lineage and review packet

CMP-014 / REVIEW RECORD

Open the reasoning.

Human review

Why this candidate appears in the shortlist

Supported comparisonThe candidate is inside the model’s applicability domain.
Uncertainty retainedThe prediction interval remains attached to the ranking.
Evidence you can inspectAssay context, preparation decisions and source records are linked.

SRScientific reviewerDecision pending · illustrative review packetAwaiting review

Built for

Head of R&DResearch leadershipData science and AI leadership

The rest of the Research & Development

Supported workflow

Diagnostics

Simulate 1%. Predict the rest.

Supported workflow

ADMET

Catch liabilities early.

Want Therapeutic running on your data?

Tell us the process or prediction you need. We'll map it to a module, a deployment and an honest timeline.

Bring your challenge.
Let’s find the right next step.

Request a Free Data Assessment