Bonito Bio and DaltonTx collaborate on agentic AI for CNS-targeted oligotherapeutics delivery
Bonito Biosciences and DaltonTx are combining large-scale ligand screening with agentic AI to tackle one of the most persistent bottlenecks in RNA medicine: delivering oligotherapeutics to the central nervous system. The two private companies announced the collaboration on May 27, pairing Bonito's giga-scale encoded ligand libraries and functional cellular selection platform with DaltonTx's adaptive AI system to design and optimize targeted delivery ligands for conjugated oligonucleotide payloads. Financial terms were not disclosed.
Bonito Biosciences, based in Waltham, Massachusetts, screens hundreds of billions of encoded ligands against complex cellular delivery phenotypes to generate proprietary datasets linking molecular structure to delivery function. The company describes its platform as generating high-dimensional information on receptor engagement and intracellular delivery — functional readouts, not binding affinity alone — across biologically relevant cell systems.
DaltonTx, based in London and founded on research from the University of Oxford, applies what it describes as an agentic AI architecture to molecular design problems. The system is designed to reason autonomously through multi-step scientific tasks, propose candidate structures, and refine predictions continuously as new experimental data are incorporated. The company was built by scientists with prior experience at AstraZeneca and Exscientia.
Together, the companies aim to establish an iterative learning loop: Bonito's wet-lab screening generates structured structure-function datasets; DaltonTx's AI ingests those data to perform structural prediction, affinity mapping, and developability analysis; the resulting ligand candidates feed back into experimental validation cycles. The initial application is CNS delivery, including novel bispecific approaches for targeted delivery to CNS cell populations.
No named drug candidates, specific disease indications, or clinical programs are referenced in the announcement. The collaboration is entirely preclinical and pre-candidate in scope. Validated delivery ligands, if produced, would subsequently need to be conjugated to specific oligotherapeutic payloads and advanced through standard IND-enabling work before any clinical entry.
The CNS delivery problem for oligonucleotides remains largely unsolved at the systemic level. GalNAc-conjugate chemistry has established liver-targeted siRNA delivery as a viable commercial modality — Alnylam Pharmaceuticals' inclisiran and givosiran are examples of approved GalNAc-conjugated products — but no equivalent validated solution exists for the CNS. That gap has drawn increasing platform-level investment from companies attempting to extend the reach of RNA medicines beyond the liver.
The Bonito-DaltonTx structure — a discovery-stage platform collaboration with no disclosed economics — is consistent with early-stage technology partnerships where IP ownership remains with each party and downstream program-specific deals are anticipated once candidates emerge. It differs from asset-licensing transactions, where upfront payments and milestone structures are standard disclosures. The absence of financial terms in this case reflects the pre-candidate nature of the work rather than an atypical deal structure for this stage.
DaltonTx's founding team background at Exscientia is notable given that Exscientia itself built an early profile through AI-assisted drug design collaborations before being acquired by Recursion Pharmaceuticals. The CNS RNA delivery space has also attracted manufacturing-focused partnerships: Apertura Gene Therapy and Viralgen recently partnered to manufacture a blood-brain barrier-crossing AAV capsid for CNS gene therapy, reflecting parallel investment in CNS access across multiple modalities. The Bonito-DaltonTx collaboration addresses the same biological access problem through a chemically distinct route — ligand-mediated receptor targeting rather than viral vector engineering.
Platform-level AI collaborations in drug discovery have proliferated, but most have focused on small molecule target engagement rather than delivery biology for nucleic acid therapeutics. The specific application of agentic AI to oligonucleotide delivery ligand design, guided by giga-scale functional screening data, represents a narrower and more technically specific problem than general-purpose AI drug discovery.
This article was generated with AI assistance and reviewed and edited by the AllSci editorial team Explore more at AllSci News: https://allsci.com/news/
Spot something wrong? Report an issue with this article
Summary
Bonito Biosciences and DaltonTx are combining large-scale ligand screening with agentic AI to tackle one of the most persistent bottlenecks in RNA medicine:...