AI-generated magnetic materials, physics-validated

150,000 AI-designed MOF candidates 10x the coverage of the existing QMOF dataset

Built on a decade-old, 40,000-crystal materials database and validated with DFT physics — now powering autonomous, AI-driven discovery for organic materials.

40K+ organic crystals in the Database Proprietary electronic-structure dataset since 2017 150,000 AI-generated magnetic MOF candidates 10x the coverage of the existing QMOF dataset DFT-validated, physics-constrained generation 40K+ organic crystals in the Database Proprietary electronic-structure dataset since 2017 150,000 AI-generated magnetic MOF candidates 10x the coverage of the existing QMOF dataset DFT-validated, physics-constrained generation

Who We Are

A decade-old materials database, now an AI discovery engine

Doublet Labs grew out of a condensed-matter research group that spent a decade building one of the largest electronic-structure databases of 3D organic crystals in existence — 40K+ crystals with electronic, magnetic, and crystallographic properties.

That database is now the proprietary training corpus behind our AI discovery pipeline, built on peer-reviewed foundations: PLOS ONE 12(2): e0171501 (2017), and Nature Physics (2021) (Geilhufe, Olsthoorn, Balatsky).

Traction

Our AI has generated 150,000 candidate magnetic MOFs, physics-validated with DFT — 10x the coverage of the existing 20,000-entry QMOF dataset, filling in magnetic-moment ranges the field hadn't reached yet.

Mission

Compressing a 20-year discovery timeline

Trillion-dollar industries — EVs, renewable energy, quantum computing — are stuck on yesterday's materials. Discovering a new functional material the traditional way takes 10–20 years. Organic and hybrid materials are the missing axis of AI-driven discovery: today's AI platforms focus almost entirely on inorganic crystals, leaving organic materials — sustainable, abundant, and environmentally friendly — largely unexplored.

Better magnets

For EVs, motors, and renewable energy — without the rare-earth bottleneck.

Superconductors & dielectrics

Organic high-Tc superconductor candidates, plus the dielectrics that pair with them.

Semiconductors

Band-gap-screened organic and metal-organic candidates from our DFT database.

Qubits & sensing

Qubit substrates and sensing applications built on magnetic MOFs.

Methods

A closed-loop autonomous discovery pipeline

01 Live

Data

Database: 40K+ organic crystals, a proprietary training corpus.

02 In progress

AI + Physics

The RENGNN neural network combined with DFT constraints generates property-targeted candidates.

03 In progress

Synthesis

CRO partners synthesize top candidates and return characterization data today, en route to an in-house autonomous lab that closes this loop end-to-end.

04 Active

Feedback Loop

Experimental results retrain the model; every cycle, the platform gets smarter.

Our moat: a proprietary database of 40K+ characterized organic crystals, physics-based AI models (the RENGNN network trained and constrained on DFT band structures), and the autonomous discovery pipeline that ties them together — an agentic AI loop that proposes, evaluates, and refines its own candidates.

See our AI capabilities in detail →

Team

Uniquely qualified expertise

Alexander Balatsky

Alexander Balatsky, PhD

Co-Founder

A decade building the Database. ~400 papers, 2 patents, APS & AAAS Fellow — brings a long-standing vision for organics in quantum technology.

Avinash Pathapati

Avinash Pathapati

Co-Founder

PhD student in AI and quantum materials, building RENGNN — the AI model behind predicting material properties.

Roadmap

From a decade of data to an automated pipeline

2017

Database established

The Organic Materials Database launched with ~40k organic compounds and their properties.

2026

AI + physics-based generation

RENGNN and DFT constraints generate and test candidates — 150k+ new MOFs, property-driven design.

2027

Scaling search and synthesis

Expanding the search space and scaling up synthesis of top candidates.

2028

Early commercialization

Client-driven material generation begins.

2030

Full pipeline automation

Finishing automation end-to-end: prediction through synthesis.

2031

Industry-scale platform

An industry-scale automated discovery platform for organic materials.

Partner With Us

Get direct access to our data and models

Need access to our materials database, AI-generated candidates, or want to explore a joint discovery project? Reach out — we reply personally within one business day.