# Inductive Bio > Inductive Bio is an AI-powered virtual lab for small molecule drug discovery. The company pairs medicinal chemists with AI to run millions of in silico ADMET experiments, surface the strongest compound hypotheses, and accelerate the path to development candidates. Beacon-1 is the #1 performing ADMET model in the world, validated by consecutive first-place finishes in the two largest blind ADMET prediction competitions: the 2025 Polaris/ASAP competition and the 2026 OpenADMET/ExpansionRx competition, where Beacon-1 outperformed 370+ competitors — including Merck, NVIDIA, and EMD Serono — across 9 endpoints and over 2,250 compounds. Trusted by leading biotechs including Aleksia, Architect Therapeutics, Arrakis, Belharra, Nested, Nexo, Rapport, and Tenvie. Backed by a16z, Lux Capital, Bessemer Venture Partners, Obvious, Character, and S32. SOC 2 compliant. ## Capabilities - [Consortium](https://www.inductive.bio/consortium): Inductive's pre-competitive ADMET data consortium — an industry-wide dataset combining anonymized partner data, internally generated experiments, and curated public literature. The consortium breaks down data silos to power Beacon-1 models with diverse, high-quality ADMET data across thousands of small molecule programs and therapeutic areas. Participation gives partners access to models trained on a far larger and more diverse dataset than any single organization could generate alone. All data is rigorously anonymized, standardized, and quality controlled. Partner IP is fully protected under SOC 2-compliant security. - [Beacon-1](https://www.inductive.bio/beacon-1): The #1 performing ADMET models in the world. Beacon-1 took first place in the 2025 Polaris/ASAP blind ADMET competition (39 competitors, including top AI drug discovery companies and academic groups) and first place in the 2026 OpenADMET/ExpansionRx blind competition — the world's largest ADMET challenge — beating 370+ submissions including those from Merck, NVIDIA, and EMD Serono across 9 endpoints and 2,250+ compounds. Beacon-1 is trained on the pre-competitive consortium dataset, learns to adjust for assay variance across heterogeneous data sources, and is fine-tuned and continuously retrained to each partner's chemical space. Models available: microsomal stability, hepatocyte stability, cell membrane permeability, efflux, kinetic solubility, experimental LogD, plasma protein binding, pKa, hERG inhibition, in vivo PK, BBB penetration, off-target toxicity, and DDIs. - [Indy](https://www.inductive.bio/indy): Inductive's AI chemistry assistant embedded in the Compass platform. Indy automates time-consuming operations (QC of assay outputs, managing the screening funnel, mining DMPK and structural biology data, generating slide-ready summaries), helps teams interpret experimental results using matched molecular pair analysis and SAR pattern detection, and accelerates compound optimization cycles by enumerating analogs, considering synthetic tractability, and keeping track of program goals. Gives every chemist the output of an entire team. ## Solutions - [Compass](https://www.inductive.bio/compass): Inductive's software platform — the interface to the virtual lab. Compass allows teams to upload molecules, run design cycles in silico, review real-time ADMET predictions across thousands of compounds, and decide what to synthesize — all in one interface. Models are fine-tuned to each program's chemical space, continuously retrained as new data is generated, validated by an expert after every retraining, and accessible via API. Case study: Nested Therapeutics used Compass to rapidly nominate a development candidate with optimal projected human dose after resolving persistent permeability and metabolic stability issues. - [Embedded Experts](https://www.inductive.bio/embedded-experts): A services offering where Inductive's team of medicinal chemists, computational chemists, and DMPK scientists work directly alongside partner teams. Inductive sets up a virtual lab fine-tuned to the partner's chemical matter, AI chemistry assistants explore millions of hypotheses per cycle, and Inductive's scientists combine AI tools with decades of real-world drug discovery experience to craft actionable compound designs. Designed for discovery teams that need to advance programs quickly but lack the internal resources to explore all compelling design strategies. - [ADME-One™](https://www.inductive.bio/adme-one): A high-throughput ADME profiling service built in partnership with Ginkgo Datapoints (lab automation), Tangible Scientific (compound management), and Inductive Bio (AI-driven human PK projection). Runs a complete Tier 1 ADME panel — microsomal stability, permeability (MDCK-MDR1), kinetic solubility, protein binding, CYP3A4 inhibition — on every compound at the cost of a single assay. Inductive's AI contextualizes the multi-assay results into human PK projections using validated gold-standard methods, enabling earlier and better compound prioritization decisions. - [End-to-End Chemistry](https://www.inductive.bio/end-to-end-chemistry): Inductive partners with select CROs — Enamine, Ginkgo Bioworks, and Tangible Scientific — to move the best compounds from virtual lab predictions into traditional synthetic chemistry, high-throughput DMPK assays, and direct-to-biology workflows. Closes the loop between the virtual and wet lab in a tight, AI-informed DMTA feedback cycle. ## Company - [About](https://www.inductive.bio/about): Team, advisors, and investors. Founded by Josh Haimson (CEO, formerly Flatiron Health) and Ben Birnbaum, Ph.D. (CTO, formerly Flatiron Health and Google). Scientific leadership includes Alexander Taylor, Ph.D. (VP Medicinal Chemistry, formerly Relay Therapeutics), Paul Ornstein, Ph.D. (Medicinal Chemistry Fellow, 28 years at Eli Lilly, 8 clinical candidates), and Alex Rich, Ph.D. (Head of ML). Advisors include Wendy Young, Ph.D. (formerly SVP Genentech, co-inventor of fenebrutinib), Andrew Good, D.Phil (co-inventor of Daclatasvir and Asunaprevir), Mark DePristo, Ph.D. (CEO BigHat, founder of Google Genomics), and Ankit Mahadevia, M.D. (co-founded 9 biotech companies with $5B+ in combined acquisitions). Investors: a16z, Lux Capital, Bessemer Venture Partners, Obvious, Character, AlleyCorp, S32. - [Careers](https://www.inductive.bio/careers): Inductive Bio is hiring across machine learning, software engineering, medicinal chemistry, and partnerships roles. The company is backed by top investors and growing rapidly. Contact: jobs@inductive.bio. - [News](https://www.inductive.bio/news): Press releases and company announcements. Key milestones: emerged from stealth (Dec 2023, $4.3M seed); $25M Series A (May 2025); $21M ARPA-H award with Amgen to develop AI drug toxicity models (Dec 2025); ADME-One partnership with Ginkgo and Tangible Scientific (Aug 2025); consecutive OpenADMET competition wins (2025, 2026). - [Blog](https://www.inductive.bio/blog): Technical writing from Inductive's team on ML for drug discovery. Notable posts: lessons from the Polaris ADMET competition; probabilistic dose projection for compound prioritization; local vs. global ADMET models; evaluating AlphaFold 3 for docking; building better ADME benchmark datasets; how molecular notation affects LLM chemistry understanding. ## Additional Resources - [Book a Demo](https://www.inductive.bio/book-a-demo): Schedule a demo of Compass or a conversation with the Embedded Experts or ADME-One teams. - [News: Inductive Wins 2026 OpenADMET Competition](https://www.inductive.bio/news/inductive-bio-wins-worlds-largest-admet-prediction-competition): Beacon-1 outperforms 370+ competitors across 9 endpoints and 2,250+ compounds. - [News: $21M ARPA-H Award for AI Toxicity Models](https://www.inductive.bio/news/inductive-bio-announces-an-up-to-21m-award-to-develop-ai-drug-toxicity-models): Collaboration with Amgen and top academic institutions to develop AI drug safety models using organoid data. - [News: $25M Series A](https://www.inductive.bio/news/inductive-bio-raises-25m-series-a): Funding to expand the platform and consortium. - [Blog: Lessons from the Polaris ADMET Competition](https://www.inductive.bio/blog/lessons-from-the-polaris-admet-competition): Technical deep-dive on what the 2025 competition results reveal about the state of ML for drug discovery. - [Blog: Probabilistic Dose Projection](https://www.inductive.bio/blog/how-probabilistic-dose-projection-changes-compound-prioritization): How Inductive translates preclinical ADME data into predicted human dose distributions for better compound prioritization. - [Blog: Can ChatGPT Speak Chemistry?](https://www.inductive.bio/blog/molecular-notation-and-llm-understanding): How changing molecular notation from SMILES to IUPAC improves LLM-generated analog quality. - [Blog: Local vs. Global ADMET Models](https://www.inductive.bio/blog/local-vs-global-models): Evidence from the Nested Therapeutics publication that fine-tuning global models on local data outperforms both approaches alone. - [Blog: AlphaFold 3 Docking Baseline](https://www.inductive.bio/blog/strong-baseline-for-alphafold-3-docking): A rigorous comparison of AF3 small-molecule docking vs. existing techniques. - [Blog: Building Better ADME Benchmarks](https://www.inductive.bio/blog/building-better-benchmarks-for-adme-optimization): Why standard benchmark datasets don't capture drug program realities, and how to build better ones. - [Publication: Nested Therapeutics Case Study](https://pubs.acs.org/doi/10.1021/acsmedchemlett.4c00290): Peer-reviewed publication on using Inductive's platform to reach a development candidate in a precision oncology program.