Expanding the universe of druggable targets - by watching proteins move.
Ensem Therapeutics, 880 Winter Street, Waltham - a 21-person biotech running a US-and-China cancer trial. Logo: company brand mark.
In drug discovery, "undruggable" is less a fact than a confession - a way of admitting that a protein hasn't been looked at closely enough. Ensem Therapeutics, a clinical-stage biotech tucked into an office park on Winter Street in Waltham, Massachusetts, was built on that distinction. Instead of treating a target protein as a single frozen crystal structure, Ensem models it as a Kinetic Ensemble: a whole population of shapes the molecule flickers through as it breathes, twists and folds. Somewhere in that motion are pockets that appear only briefly - cryptic and allosteric sites a static snapshot would never show. Find one, fit a small molecule into it, and a target long written off suddenly becomes tractable.
That is the whole thesis, and it is not just marketing. Founded in 2021 and incubated by the healthcare investment firm CBC Group, Ensem has already turned its platform into two molecules dosed in human patients - one of them advanced by a global pharmaceutical partner, the other carrying an FDA Fast Track designation. For a company of roughly 21 people, that is an unusual amount of clinical output, and it is the clearest evidence the approach produces more than slides.
Ensem designs small-molecule precision medicines for cancer, aimed squarely at targets that conventional structure-based design struggles to reach. The engine is the Kinetic Ensemble platform, which couples three things that usually live in separate buildings: AI and machine-learning models, physics-based molecular-dynamics simulations, and hands-on experimental biophysics.
The simulations and AI propose where hidden pockets might open and how a candidate molecule could bind. Crucially, those predictions do not stand alone. They are checked against a stack of experimental methods - x-ray crystallography, cryo-EM, HDX-MS and NMR dynamics - so that a computational hunch becomes a validated binding site rather than a guess.
That validation loop is the point. Plenty of companies now say the words "AI drug discovery." Ensem's distinction is that its predictions are tethered to wet-lab reality at every step, which is what lets it chase hot-spot mapping, transient pockets and metastable conformational states with some confidence.
The output is a pipeline of small molecules engineered to be selective - hitting a mutant or disease-driving form of a protein while sparing the healthy version. In oncology, that selectivity is not a luxury. It is the difference between a drug patients can tolerate and one they cannot.
A potential first- and best-in-class allosteric, pan-mutant-selective PI3Kα inhibitor and degrader for PIK3CA-mutant, HR+/HER2- advanced breast cancer. Designed to inhibit and destroy mutant PI3Kα while sparing the wild-type protein - aiming for efficacy with better tolerability. FDA Fast Track granted; dosing in the US (June 2025) and China (February 2026).
A highly potent CDK2 inhibitor with roughly 100x selectivity over other CDK-family kinases, designed to make previously unexplored interactions inside the CDK2 ATP binding pocket. Advanced into a first-in-human study for solid tumors through partner BeiGene.
The discovery platform itself - AI/ML plus molecular dynamics plus experimental biophysics - that generated both clinical candidates and continues to feed the pipeline against difficult-to-drug oncology targets.
Pipeline stages are illustrative of publicly disclosed clinical status, not precise progress percentages.
Two of the biggest families in cancer biology - the PI3K pathway and the CDK cell-cycle kinases - are notoriously hard to drug well. The active sites look alike across related proteins, so molecules that block the disease-driving version tend to hit healthy ones too, and patients pay for it in side effects.
Ensem's answer is to aim somewhere other than the obvious active site. By modeling proteins in motion, the Kinetic Ensemble platform surfaces allosteric and cryptic pockets - places that are more distinctive between the mutant and the wild-type form, and therefore easier to target selectively.
ETX-636 is the clearest illustration. Rather than competing at the crowded ATP site, it slots into a specific allosteric pocket on p110α, the catalytic subunit of PI3Kα, and is engineered to selectively inhibit multiple activating mutant forms while leaving the wild-type enzyme alone. It does not only block the mutant; it degrades it.
Against pure in-silico players, Ensem's edge is the wet lab. Against traditional medicinal-chemistry shops, its edge is the physics-informed AI. The company sits in the overlap - close enough to computation to move fast, close enough to the bench to trust its own predictions.
It competes in a busy field. Platform biotechs such as Relay Therapeutics, Nimbus and Scorpion are also chasing allosteric and difficult-to-drug targets, and several larger companies are developing mutant-selective PI3Kα inhibitors of their own. Ensem's wager is that its motion-first view of proteins yields pockets - and therefore drugs - the others miss.
A venture-backed clinical-stage biopharma. The Kinetic Ensemble platform generates proprietary drug candidates; value is realized through partnerships and licensing - as with BeiGene on CDK2 - and future clinical milestones, commercialization or acquisition.
Ultimately cancer patients enrolled in trials across the US and China. Near-term stakeholders include pharma partners, clinical investigators at leading cancer centers, and the investors funding the pipeline.
Ensem lives at the intersection of AI drug discovery and small-molecule precision oncology - a segment where a defensible platform, not headcount, is the moat.
Ensem is led by co-founder, President and CEO Shengfang Jin, Ph.D., who brings more than 25 years in oncology and therapeutics development. She holds a Ph.D. in molecular biology and microbiology from Tufts, trained as an NIH postdoctoral fellow in oncology and tumor immunology at Harvard Medical School, and previously served as Vice President of Discovery Biology at Editas Medicine, where her teams worked on advanced CRISPR gene and cell therapies.
The wider team is deliberately cross-disciplinary - computer-aided drug design enhanced by AI, alongside biophysics and structure biology, medicinal chemistry, and DMPK - with senior scientific and clinical leadership including a Chief Scientific Officer and Chief Medical Officer. It is a small organization built so that computational and experimental drug discovery sit in the same room rather than across a handoff.
Deep-learning models and molecular-dynamics simulations that predict cryptic pockets and binding modes.
Cryo-EM, x-ray crystallography, HDX-MS and NMR dynamics validate predictions against experimental reality.
Medicinal chemistry and drug-metabolism expertise turn validated pockets into clinical-quality molecules.
Established in Waltham and incubated by CBC Group to build small-molecule medicines for difficult-to-drug targets.
Closed a Series A led by GGV Capital, with CBC Group, Pavilion Capital, Cenova Capital and Mitsui & Co. Global Investment participating.
Co-founder Shengfang Jin promoted to lead the company.
ETX-197/BG-68501 advanced into trials for solid tumors via BeiGene; preclinical CDK2 data presented at the San Antonio Breast Cancer Symposium.
First US patient dosed in June; FDA Fast Track designation granted for advanced breast cancer in October.
First patient dosed at Fudan University Shanghai Cancer Center; proof-of-concept data expected in the second half of the year.
It is a clinical-stage biotech that designs small-molecule precision cancer medicines against difficult-to-drug protein targets, using its AI-and-physics-based Kinetic Ensemble platform.
A suite of AI/ML, molecular-dynamics simulation and experimental biophysics technologies that model proteins as moving ensembles to reveal cryptic and allosteric binding pockets missed by static structures.
ETX-636, an allosteric pan-mutant-selective PI3Kα inhibitor and degrader for breast cancer, and ETX-197/BG-68501, a selective CDK2 inhibitor partnered with BeiGene.
A $67 million Series A led by GGV Capital, with investors including CBC Group, Pavilion Capital, Cenova Capital and Mitsui & Co. Global Investment.
Its headquarters is at 880 Winter Street, Waltham, Massachusetts, with a team of roughly 21 people.