Digital biology & drug discovery · Gut microbiome AI
Anto Biosciences
A San Francisco startup building a multimodal foundation model for the human gut microbiome, aimed at predicting which patients a drug will fail in before it reaches a clinical trial. Founded in 2025, it is a Y Combinator Fall 2025 company that has announced a seed round but has not disclosed the amount.
A billion patients, one hidden variable
Anto Biosciences opens its pitch with a specific failure. According to the company's own account, posted on Y Combinator's site, a cancer drug worked in trials in China and then failed in the United States, the same molecule producing different outcomes in different patients: “We caught a cancer drug that worked in China but failed in U.S. trials. Same molecule, different microbiomes. $1B lost.”
The company says more than a billion people currently take medications whose success depends on the bacteria living in their gut. Gut microbes can metabolize a drug before it reaches its target, convert an inactive compound into an active one, turn a drug into something toxic, or change how the immune system tolerates it. Pharmaceutical research and development, the company argues, still treats this variation as background noise rather than a variable that can be measured and designed around.
Two second time founders
Anto Biosciences was founded in 2025 by Arvid E. Gollwitzer and David de Gruijl, according to the company and multiple investor and directory profiles. Both describe themselves as second time founders. Gollwitzer is a graduate researcher in the Traverso Lab at the Massachusetts Institute of Technology and a researcher at the Broad Institute of MIT and Harvard, per his Forbes Business Council profile and the company's team page. He studied at ETH Zurich, MIT, and Harvard Medical School as a Swiss Study Foundation fellow, and he teaches Clinical Genomics at ETH Zurich. He previously co-founded NextGuide, described by the company as an AI product for visually impaired users.
De Gruijl was a researcher in the Gastroenterology Department at Brigham and Women's Hospital and Harvard Medical School, where he built in vitro gut models to study microbial interactions, according to the company's team page. Before Anto, he was Chief Technology Officer of Nook Biosciences, described as a drug microbiota company, from 2024 to 2025, per the investor research site DealBio. Earlier he worked in data science at Janssen Pharmaceuticals, part of Johnson & Johnson, after engineering degrees at Delft University of Technology. Announcing the company on LinkedIn, he described trying to rebuild the digestive tract in a lab: “Our models predict how gut microbes influence digestion and drug response across diverse populations.” He added: “If you’re in drug development and suspect the digestive tract might be affecting your products: you’re probably right.”
Sparsification, from genome archives to the gut
Gollwitzer's earlier research is in high performance genomics. He is credited as a co-developer of GenStore and MegIS, in storage processing systems for genome sequence analysis built in Onur Mutlu's SAFARI research group at ETH Zurich, and as first author of MetaTrinity, a metagenomic classifier. The company describes his central technique as computational sparsification: training a system to ignore most of a dataset and focus computation on the fraction that predicts an outcome. The company's website applies the same idea to the gut: “We remove noise and low-quality data through sparsification, leaving only the ~1% that actually carries predictive power.”
Anto's listed research includes two papers presented at the NeurIPS 2025 AI for Science workshop with MIT's Giovanni Traverso as a co-author: MetaOmics 10T, described as a foundational dataset for causal modeling of microbial ecosystems, and a paper on hierarchical control of therapeutic resistance in cancer treatment. De Gruijl is a co-author on a separate, peer reviewed paper, "Emerging Tools and Technologies for Microbiome-Aware Drug Development," published in Clinical Pharmacology & Therapeutics in 2025 with H. J. Haiser. The investor research site DealBio credits the company's filtering method with increasing the usable share of public sequence data from 5% to 40%.
A seed round with a widening investor list
Anto joined Y Combinator's Fall 2025 batch, with Nicolas Dessaigne listed as its primary partner, according to the company's Y Combinator profile. Announcements of a seed round have surfaced from more than one backer. A LinkedIn post from 007 Venture Partners describes the deal as that firm's sixth investment and names Y Combinator, Massive Tech Ventures, Ritual Capital, RRE Ventures, and Standard Partners as co-investors, alongside 007 itself. Separately, the deal intelligence sites Dealroom and Preqin report a seed round tied to Tekedia Capital and dated to June 2026, and Preqin lists gumi Inc. as a current investor. No source reviewed states a round total, a valuation, or a single closing date, and a search of SEC EDGAR company filings under the name Anto turned up no Form D for Anto Biosciences, Inc. as of September 25, 2026.
What is proven, and what is still claimed
| Evidence | What the record shows | Source type |
|---|---|---|
| Regulatory and clinical record | No FDA 510(k), PMA, or registered clinical trial found under Anto Biosciences or Anto Bio in FDA and ClinicalTrials.gov searches as of Sep 25, 2026. The company builds a computational prediction platform rather than a therapeutic, so an absence of trial records is not itself unusual at this stage. | Not found |
| Federal grants | No NIH or NSF award naming Anto Biosciences was found. NSF records that surfaced under a fuzzy company name search belong to unrelated companies that also include the word Biosciences in their names. | Not found |
| Preprints and workshop papers | MetaOmics 10T and a paper on hierarchical control of therapeutic resistance in cancer treatment were presented at the NeurIPS 2025 AI for Science workshop, both crediting Gollwitzer with MIT's Giovanni Traverso. Earlier preprints, GenStore, MegIS, and MetaTrinity, are posted to arXiv and cited by the company. | Public record |
| Peer reviewed journal paper | "Emerging Tools and Technologies for Microbiome-Aware Drug Development," by H. J. Haiser and David de Gruijl, Clinical Pharmacology & Therapeutics, 2025, doi 10.1002/cpt.70026, indexed on PubMed. | Public record |
| Y Combinator participation | Listed on Y Combinator's site as a Fall 2025 batch company, status active, team size 4, primary partner Nicolas Dessaigne. | YC data |
| Team size | Y Combinator lists a team of 4. The research site Startup Intros describes the company as operating with a core team of two full time employees. The two figures have not been reconciled by any source reviewed. | Differs across sources |
| Seed funding | A 007 Venture Partners LinkedIn post names Y Combinator, Massive Tech Ventures, Ritual Capital, RRE Ventures, and Standard Partners as co-investors. Separately, Dealroom and Preqin report a seed round tied to Tekedia Capital, and Preqin lists gumi Inc. as a current investor. No dollar amount is confirmed by any source, and no SEC Form D for the company was found. | No amount confirmed |
| Named pharma partners and customers | The company says it is working with pharma research and development teams to simulate drug microbiome interactions, but no partner or customer is named in any source reviewed. | Company-stated |
Read plainly: Anto has a real research pedigree, a peer reviewed paper and preprints that predate the company, and a Y Combinator listing confirming it is an active startup. Missing is anything showing the platform working on an outside party's drug, a named pharma partner, or a clear accounting of how much money it has raised. Two investor lists reported months apart, with no Form D to settle them, is the clearest gap.
What to watch
- A named pharma partnership, pilot result, or published case study using the Darwin model series on a real drug.
- A Form D or other public disclosure that states the seed round's actual size.
- Reconciliation of the two different team size figures and the two different investor lists.
- Peer reviewed validation of the Darwin models themselves, beyond the foundational sparsification and dataset preprints.
In their words
“We caught a cancer drug that worked in China but failed in U.S. trials. Same molecule, different microbiomes. $1B lost.”
Arvid E. Gollwitzer and David de Gruijl, Y Combinator launch post, 2025 · Company
“We remove noise and low-quality data through sparsification, leaving only the ~1% that actually carries predictive power.”
Anto Biosciences, company website, accessed Sep 25, 2026 · Company
“We predict drug toxicity and efficacy across diverse populations and fix drugs so they work for everyone.”
Anto Biosciences, Y Combinator company profile, 2025 · Company
“Our models predict how gut microbes influence digestion and drug response across diverse populations.”
David de Gruijl, co-founder, LinkedIn, Oct 2025 · Company
“If you’re in drug development and suspect the digestive tract might be affecting your products: you’re probably right.”
David de Gruijl, co-founder, LinkedIn, Oct 2025 · Company
“If you can simulate drug response across geographies and diets before clinical trials, failure becomes an optimisation proble, not a billion-dollar gamble.”
007 Venture Partners, investment announcement, LinkedIn, 2026 · Investor
“Two patients can take the same drug at the same dosage and experience entirely different results because their microbial communities differ.”
Dealroom, investor note on Tekedia Capital's investment, Jun 2026 · Independent
Related companies
Sources
- Public recordPubMed record, Emerging Tools and Technologies for Microbiome-Aware Drug Development
- Public recordSEC EDGAR company search, companies matching "Anto," Form D
- YC dataAnto Biosciences: A Foundation Model for Microbial Communities
- IndependentArvid Ernst Gollwitzer, Founder, Anto Biosciences
- IndependentAnto Biosciences investor profile
- IndependentAnto Biosciences, Inc. asset profile
- IndependentTekedia Capital invests in Anto Biosciences to make the gut microbiome computable
- IndependentAnto Biosciences company profile
- IndependentAnto Biosciences funding profile
- IndependentAnto Biosciences Inc, company profile
- InvestorAnto Biosciences Raises Seed Round for Microbiome Modeling
- CompanyBig News: Arvid and I are building Anto Biosciences
- CompanyAnto Biosciences homepage, team page, research page, and publications
- CompanyFoundation Models for Microbial Communities
- CompanyAI-controlled Metagenomics: High-Throughput Taxonomic Profiling for Clinical Pathogen Detection
Profile researched and written by Healthcare Discovery. Last updated September 29, 2026.
