🔗 Blog: manifold.bio/news/mber-2-sca…

Excited to be an early partner in Anthropic's Life Sciences Verification Program

We now have speakers. Accepting research submissions for our 2026 NeurIPS workshop in Paris, France on data infrastructure + benchmarks for scientific AI. The workshop focuses on: (1) how scientific data should be organized so models and agents can use it (2) how evaluations should determine where frontier models are reliable Submit here: aidar-workshop.github.io/202…

Anthropic shipped Mythos 5.1 today, leading the ProteinGym benchmark at 49.3% rank correlation against real lab measurements. Two weeks ago they published Claude running autonomous protein binder design campaigns at a 27% hit rate versus the 10-15% that's typical today. Both results are impressive, but both point to a problem most people aren't (yet) seeing. Start with the shape of that ProteinGym chart: Mythos 5.1 - 49.3% Opus 5 - 47.7% Mythos 5 - 45.8% Gemini 3.1 Pro - 37.0% Sonnet 5 - 36.6% GPT-5.6 - 35.5% Six frontier general-purpose models inside a 14-point band. None of them dedicated protein models. All trained on roughly the same public corpus. That's not anyone achieving a moat; it's a a floor rising. What ProteinGym actually measures: ~217 deep mutational scanning assays covering millions of variants, where every one of them is an experiment somebody already ran and published. Predicting results that exist is a different problem from generating results that don't. The binder campaign has the same shape, and Anthropic says so in its own technical report. Co-folding confidence turned out to be a useful filter but not a guarantee, and "experimental screening remains the only way to learn" which targets a campaign actually succeeded on. Adaptyv, who ran the wet lab, called it an open-loop experiment and said the next step is closing the loop. You can't close that loop at 30 designs per target with a multi-week CRO turnaround. You get one shot and no statistical power to learn anything from it. In contrast – last March, @ManifoldBio and @NVIDIA tested 1,000,000 designs against 127 targets in a single multiplexed experiment, measuring over 100 million protein-protein interactions. Roughly 750x the design count of the Anthropic campaign, in one run. (disclosure: I'm on Manifold's board, and an investor.) The model they were validating in that study was Proteina-Complexa, which is one of the ten open-source design models Claude used in the Anthropic campaign. Design is converging on a shared toolkit, and ... commoditizing. Measurement is not. Which brings me to the ladder below. Every public benchmark, and both of Anthropic's recent headline results, live near the bottom of it. Expression. Binding. In vitro affinity. There is no ProteinGym for biodistribution and for what actually translates into human biology (and into therapeutics that work). Not because nobody wants one, but because the data doesn't exist at benchmark scale. Someone has to generate it, at billion-scale. Whoever does has the data that matters to train the models that will actually make a difference. My bet is that the first to million-scale will be the first to billion-scale.

Manifold Bio is heading to @RosettaCommons Summer RosettaCon 2026 next week. Erik Swanson and Mike Nichols will present the EpiTome, our proteome-scale effort to design binders to every addressable epitope on every human cell-surface receptor. We’ll talk about highlights from our latest million-scale de novo antibody design and validation campaigns. Our data now exceeds the number of antibody-antigen interaction complexes in the PDB by more than an order of magnitude. We're pushing the boundary of data scale in protein design, closing the loop between design and validation at a pace and density the field has never seen. DM to connect with Erik or Mike. 📅 August 6, 2026, 1:30pm PT 📍 Seattle, WA

Thank you to the @decodingbio team for having me at the AI x Bio summit at the NYSE last week. Great panel chaired by @ElliotHershberg on how new technologies can create both new types of medicines and new kinds of companies.

𝐔.𝐒. 𝐒𝐜𝐢𝐞𝐧𝐜𝐞 𝐒𝐮𝐦𝐦𝐢𝐭: Anthropic called out @ManifoldBio as a key feedback provider when they announced Claude for life sciences recently. CEO @glebkuz explains, and updates us on the brain shuttle program Full video: biotechtv.com/post/manifold-…

New drug-creation technologies not only open up new medicines, but also the kinds of companies that can be built. Looking forward to digging into this on a panel at the Decoding Bio AI x Bio Summit this week with @BenjamineYLiu , @SurgeBiswas , and @zavaindar , moderated by @ElliotHershberg . For @ManifoldBio, our ability to run million-scale protein design and in vivo testing has already let us establish a beachhead in tissue-targeted medicines. That space holds hundreds, if not thousands, of potential new drugs, and it's red hot in pharma right now. Our first deal with Roche (up to $2B / $55M upfront) is the template for our core strategy: partner around specific products, while we keep investing in and reusing the underlying targeting tech across other partnerships and our own programs. That lets us lean into the platform's compounding advantage and, over time, take on more of the product creation ourselves. Looking forward to sharing more Thursday 7/23 at 2:10pm at the @NYSE.

Getting the right drug to the right tissue is one of the hardest problems in medicine. @ManifoldBio is presenting data on how we're solving it at the Barriers of the CNS Gordon Research Conference next week. Supriya Ravichandran will present our poster on ultra high-throughput in vivo screening for CNS shuttle discovery. Screening hundreds of thousands of in silico-designed molecules simultaneously in vivo, we identify novel receptor portals and tissue-targeting shuttles with differentiated biodistribution profiles. Massively multiplexed in vivo screening is how we turn drug localization from a bottleneck into a solvable design problem, with the same engine extending across tissues and indications. Attending? DM us to connect with Supriya or Marshall Case. 👋 📅 June 14-19, 2026 | New London, NH 📍 grc.org/barriers-of-the-cns-…

Last week, @Anthropic brought together a select group of founders for their Founders’ Lab during #BOSTechWeek. Super to have our Co-Founder & CEO, @glebkuz , be selected to share how @ManifoldBio is doing science in the new era of agents!

We’re excited to join @AnthropicAI for the Founders’ Lab: AI for Life Sciences during #BOSTechWeek tomorrow May 27th in Boston. @ManifoldBio 's Co-Founder & CEO, @glebkuz , will speak on a panel with other AI x Bio leaders about how Manifold is building the Virtual Organism and the agentic infrastructure enabling it. 📍 Boston 🗓 Wednesday, May 27 ⏰ 1:30–5 PM ET Limited-capacity event; approval required to attend. Event Link: partiful.com/e/gKBL52v6jIOhz…

If you're attending @PEGSboston, don't miss the talk by our Co-founder and CTO, @PierceOgdenJ, on Manifold Bio's approach to one of the hardest problems in drug development: getting the right drug to the right tissue. Pierce will present our platform, which connects AI-driven protein design to functional readouts in living systems. Using massively parallel molecular multiplexing, Manifold can evaluate hundreds of thousands of designs simultaneously in any biological context, including in vivo. Our multiplexed in vivo screening platform allows simultaneous assessment of tissue distribution across hundreds of receptor candidates in a single study. This enables programmable, selective delivery of diverse therapeutics across the body, driving a pipeline of tissue-targeted medicines. Directly after his talk, Pierce will join a panel on de novo biologics design chaired by @SurgeBiswas (@nablabio), alongside Prashanth Vishwanath (@TakedaPharma) and Maria Wendt (@sanofi). 📅 Friday May 15, 2026 | 10:55 am ET 📍 PEGS Boston - pegsummit.com

Heading to #TIDESUSA? Don't miss the talk by our Co-Founder and CTO, @PierceOgdenJ. Current oligonucleotide delivery technologies are not optimized for diverse tissues and therapeutic product profiles. Pierce will present data from @ManifoldBio's multiplexed in vivo screening platform, which we used to test thousands of AI-designed protein shuttle candidates against hundreds of novel receptors directly in living systems, revealing diverse biodistribution profiles across brain and peripheral tissues and identifying new conjugate partners for improved oligonucleotide delivery. 📅 Wednesday May 13, 2026 | 2:30pm ET 📍 TIDES - informaconnect.com/tides