Valyra
AI-Powered Drug Discovery Platform

Design better drug
candidates, faster.

Valyra connects molecular design, ML prediction, and candidate review into a unified scientific workbench — purpose-built for drug discovery teams.

Launch App
Valyra — Design Workbench
Compound
SMILES
Solubility
LogP
Drug-likeness
Status
CPD-0042
CC(C)Cc1ccc...
−1.8
3.2
0.91
pass
CPD-0043
c1ccc(cc1)C...
−2.1
4.1
0.78
review
CPD-0044
O=C(O)Cc1cc...
−0.9
1.8
0.94
pass
CPD-0045
CN1CCC(CC1)...
−3.4
5.2
0.44
fail
CPD-0046
COc1ccc(cc1...
−1.5
2.7
0.88
pass
3 / 5
passing

Features

Everything a discovery team needs in one place

Molecular Design Workbench

Draw, enumerate, and score candidates in a unified environment. Compare multi-result sets side by side and aggregate findings across experiments.

SMILESSDFMulti-set comparison

ML Model Registry

Connect external model providers and run predictions at scale. Register, activate, and route to any provider that implements the standard API contract.

Provider registryPrediction routingBullMQ queues

Candidate Triage

Review, score, and make go/no-go decisions on compound series. Campaign management, AI pre-review, and one-click export to downstream workflows.

Campaign managementAI pre-reviewStructured decisions

AI Assistant

Conversational interface for queries, data exploration, and workflow execution. Ask questions about your compound library or kick off multi-step agent runs.

Chat interfaceAgent runsWorkflow automation

How it works

From provider setup to decision in three steps

1
01

Connect your providers

Register model providers, tool providers, and compound databases via URL. Valyra discovers capabilities automatically — no configuration files required.

2
02

Design and evaluate candidates

Build compound series in the Design Workbench, run ML predictions in bulk, execute docking and simulation tools, and compare results across experiments.

3
03

Triage and export decisions

Bring candidates into a Triage campaign, apply scoring criteria, record go/no-go decisions with rationale, and export structured data to your downstream systems.

Built for science

Not a generic platform. Built for discovery.

Valyra was designed with computational chemists and medicinal researchers in mind. Every decision — from data model to UI — reflects how drug discovery teams actually work.

Drug discovery-specific

  • SMILES, SDF, and substructure-native data model
  • Similarity and substructure search across compound libraries
  • Docking viewer with protein-ligand 3D visualization
  • MD trajectory playback (XTC, DCD, AMBER NC)

ML-ready by design

  • External model provider registry — bring your own models
  • Prediction routing and async job queues
  • Tool provider system for docking, conformer generation, and more
  • Result set comparison and cross-workbench aggregation

Enterprise-ready

  • Role-based access control and group management
  • Auth0 / Okta SSO integration
  • PostgreSQL with pgvector for compound storage and search
  • Deploy on Vercel, AWS, or on-premises

Ready to see Valyra in action?

Get early access or launch the app now.

Launch App