Frameshifts with Benjamin Arya

This week, my X feed exploded with news that Anthropic had released hardware standards that allow AI agents to operate lab equipment directly. Their new “Model Hardware Standard” or MHS, built in collaboration with the Howard Hughes Medical Institute, allows an LLM to operate microscopes, liquid handlers, robotic arms and dozens of other instruments that scientists use to conduct experiments and to build the future.Until now, there has been no standard way to connect an LLM to a physical instrument. Every time you want to connect two pieces of equipment together, you need to create a bespoke integration that requires some grad student to write custom software. And grad students are busy. So when a lab wants to increase throughput, the incentive structure is usually to just hire more grad students or pressure encourage everyone in the lab to work longer hours. And scientists already work much longer hours than other professions. The standard informal working schedule at many elite labs across the US and China is something close to 996, often for mediocre salaries in the range of $35-70k.Of course, most people know that it’s silly for humanity to send our best and brightest to spend the better half of a decade getting PhDs in advanced bioengineering just to spend most of their time as manual labour meat sacks moving microliters of liquids between plastic tubes by hand.The bottleneck in biology has always been the ability to perform rapid iteration experiments to collect real world data. Intellect has been abundant in science since the 1950s, and with the advent of super-intelligent models, it is already becoming commoditized. These days, what distinguishes Earth’s most impactful and highest throughput labs is largely prestige-driven meat sack talent acquisition, work ethic, pipetting technique and ability to fund expensive experiments.This week’s episode with Lucas Mair is all about what the world will look like when we realise that robots exist and start building accessible, intuitive software to allow them to perform closed-loop laboratory experiments.Lucas is a very German MD-PhD who took an unusual route into biology. Computer science first, then medicine, then neural engineering at TU Munich, and before any of that, building autonomous drone swarms for search and rescue. He thinks like a control theorist, and when a control theorist looks at how a modern biology lab actually runs, he doesn't see cutting-edge science. He sees a factory floor from 1975.Lucas will tell you that most labs already have an Opentrons or something similar gathering dust in the corner. My own lab had one that I have never seen anyone use.Grants pay for reagents and salaries, not for engineers to automate your protocols. Grad students are cheaper than robots, and publishing pressure means the three months you'd spend automating a workflow is three months you're not putting out papers, which is the equivalent of suicide by irrelevancy for many high-profile labs. Thus, the rational move, for every individual scientist and every individual lab, is to just keep pipetting. And so that’s what everyone does.Which is insane, because the robots already exist. We've had them for years. What we've been missing is the thing that turns "here's what I want to test" into machines actually doing it.The Anthropic news is a bit of a preview into what that might look like. But the reality of the current tech is a mixed picture. For example, scientists at Carnegie Mellon were able to use an AI agent with MHS to orchestrate a liquid handler, plate reader, robotic arm and monitoring cameras to get serial-dilution dose-response experiments done 3x faster.Meanwhile at Genentech, humans had to intervene when Claude used MHS to coordinate a BCA protein assay because the model mistook errors caused by bubbles for a software failure, and then responded in a way that produced yet more bubbles.Lucas Mair looks at biology as a control and engineering problem. You have a complex biological system, and you want to move it from State A to State B (e.g. diseased to healthy). But the problem is there is an unimaginably large search space of possibilities. For example, there are 2020 or 104,857,600,000,000,000,000,000,000 ways of making a 20 amino acid protein sequence.Most of these sequences are useless but a few of them might help make humans immortal.One more thing to understand before you listen to the full conversation…If you’ve spent much time at all on the tech bro side of X, you’ll know about the scaling laws of large language models. Roughly speaking, intelligence scales linearly with the log of compute (or data). Thus to improve your model, you need more data.But most of the recent improvement in LLM performance across disciplines hasn’t been in training larger language models. Rather, the most noticeable advances have been from “reinforcement learning from verifiable rewards” or RLVR. Basically a form of reinforcement learning where the reward you give the model can be checked objectively rather than judged subjectively by humans. After all, if you want to build super-intelligent systems, then at some point, even the smartest humans are going to be useless in teaching them anything. RLVR is really easy to do in fields like math and coding because the correct answer is already known and you can check if the program passed the tests. But how do you do this in biology? You can’t just read all the papers because most biological data isn’t published and the results of many peer-reviewed papers can’t be reproduced by other labs (Cc: the reproducibility crisis in science).You don’t need more text tokens. You need more ground truth data (i.e. the actual underlying molecular biology) aimed at the model’s blind spots. The jagged frontier of AI progress is more like a recursive Sierpiński triangle. Each field, subfield and sub-sub-field of human endeavor has their own jagged frontiers that would benefit from RLVR.If you can generate data aimed at places where the model is uncertain or wrong, using real experiments, the efficiency gain can be enormous. Now imagine running that loop in biology, on its own. The model would propose the hypothesis, plan the experiment, dispatch agents to run the lab equipment and collect real measurements. The data then would get processed to generate the next hypothesis. This is a system that learns by doing, using verifiable, falsifiable predictions.It’s the same technique that helped AI vibe coding go from being an unusable joke to the default way that 90% of today’s code is written, including at frontier labs. Once you collect enough data on the model’s biological blind spots, you cross some invisible threshold of reliability that makes your AI model generally useful in autonomously solving biological engineering problems, inventing drugs, running experiments, and, eventually, agentically building biotech companies.Closed-loop automated labs could put real experimental biology in the hands of far more people than the credentialed few who have obtained their pipette licenses.One day, you might be able to prompt a model to fix your body, cure your cold or build your dog a cancer vaccine.Or maybe you already can, but just haven’t tried.Watch on YouTube. Listen on Apple Podcasts or Spotify.GUEST INFORMATION:* Lucas Mair, Co-Founder & CTO, ProviGenAI* ProviGenAI* Model Hardware Standard (Anthropic)* X (Twitter)* LinkedInCONNECT WITH US:* Website* Substack* YouTube* X (Twitter)* LinkedIn Get full access to Frameshifts with Benjamin Arya at frameshifts.bio/subscribe

On this show, we’ve talked a lot about gene editors. These are little molecular machines derived from or inspired by the Nobel-prize winning CRISPR system that put gene editing on the radar of the general public.Once these editors get into a cell, they can cut out DNA, change the DNA sequence, insert entire copies of genes or delete faulty genes that cause illness. Think of them as tools to cut, paste, backspace and overwrite the code of life.But the primary challenge with these gene editors, besides off-target edits and low-but-steadily-improving editing efficiencies, is that they need to get into the cell to be able to do anything. Getting gene editors into cells is an entire field of biomedical engineering called gene delivery. And in comparison to the rapidly flourishing field of gene editor tech dev, the field of gene delivery is both under-appreciated and relatively stagnant.And that’s a massive problem.Because without effective delivery vectors, gene editors are limited to a very small number of conditions for which our delivery vectors happen to be okay at. Current clinical successes are concentrated in situations where delivery is unusually favorable:* Liver diseases, where lipid nanoparticles can efficiently deliver cargo to hepatocytes after IV injection* Blood disorders and some hematological cancers, where cells can be removed from the patient, genetically modified ex vivo using lentiviruses and then re-injected into the patient (e.g. HSC therapy for sickle cell anemia or CAR-T therapy for leukemia)* Diseases that only require a very small percentage of cells to be edited in order to be functionally cured* A limited number of tissues, such as the retina or CNS where viral vectors can achieve therapeutically meaningful gene transferThat's the ceiling today. But if we crack delivery, the potential of this field is disgustingly profound.Gene Delivery Vector + Gene Editor = Gene TherapyIf humanity can solve both sides of this equation, we can, literally, cure anything. You have a genetic disease? Here’s an injection that gives all of your cells the healthy copy of the corrupted gene you inherited from your parents.You’re prone to cardiovascular disease? Here’s an injection that gives you new human-designed molecular machines that clear out your clogged arteries.You have a family history of cancer? Here’s an injection that replaces your genome maintenance and DNA machinery with engineered supernatural versions that can protect your cells from cancer-causing mutations for at least 1000 years.Worried about keeping up in a world of artificial superintelligence? Here's an injection that carries the morphological instructions for a population of your cortical neurons to rewire themselves for compatibility with high-bandwidth brain-machine interfaces, connecting you directly into Dario or Sam’s frontier models.Now, I know a lot has to go right for humanity to achieve mastery over the code of life and transcend the limits imposed by “fit enough to have offspring” evolutionary biology.So I’ll get my head out of the clouds for a moment.Currently, the most clinically successful delivery vehicle in the emerging field of gene therapy is a family of viral variants that we call Adeno-Associated Viruses, or AAVs.AAVs are DNA viruses that scientists have hijacked to make little viral couriers that carry a working gene into your cells. Within this group are many serotypes and engineered capsid variants, each with tropism for specific organs and tissue types. Significantly oversimplifying: AAV9 can penetrate the brain, AAV8 is strongly liver-tropic and other capsids are tuned for tissues like skeletal muscle or retina.But somewhere between a third and two-thirds of people already have antibodies against AAVs, left over from ordinary childhood infections with the wild versions. If you’re one of them, a gene therapy that could save your life may simply not be available to you, because your immune system will neutralize the courier before it arrives. And even if you’re in the clear, most AAV therapies work exactly once. The first dose teaches your body to recognize the vector, so a second one gets destroyed the moment it touches your immune system. For a whole field built on delivering genes into people, the delivery itself is one of the deepest unsolved problems there is.Logan Thrasher Collins has spent his career on that problem. As a PhD student in biomedical engineering at Washington University in St. Louis, he invented a new gene-delivery modality called vaultAAV, which does something a bit strange: it hides the AAV inside a protein vault. Wow. Incredible.But what the hell is a protein vault?Vaults are odd, barrel-shaped organelles our own cells churn out in huge numbers, and nobody is entirely sure what they’re for. What Logan realized is that because the body recognizes a protein vault as self, they act as immunologically invisible Trojan horses.Cells take them up, the vault disassembles in the acidifying endosome, and the AAV escapes to deliver its DNA to the nucleus, all without the immune system ever getting a look at the AAV courier hiding inside.In this episode, Logan walks us through his entire scientific journey, from designing an aggregating antimicrobial peptide as a high school science fair project (which won the top prize at ISEF and got a minor planet named after him), to the insight that connected the protein vault literature with the AAV field’s antibody problem. We discuss the full landscape of gene delivery: AAVs, adenoviruses, LNPs, virus-like particles, and why the field’s relationship with each vector is driven more by reputation and vibes than by rigorous analysis of dosing and safety data.Logan also explains why vaultAAVs may enable something no other approach can: redosability. Currently, AAV capsids are engineered with surface mutations to escape antibodies, but each new variant can only be used once before the immune system learns the new epitope. Vault-shielded AAVs, because their outer surface is self, could theoretically be administered multiple times, allowing lower doses, reduced toxicity and lower manufacturing costs.Logan has since taken the vaultAAV work out of the lab and into a company. He co-founded Cathedral Therapeutics, a gene-delivery engineering company, alongside Prof. David Curiel, MD, PhD at Washington University, to build exactly this kind of stealth delivery system.In this episode, we also get into:* How 30 to 60% of people are locked out of AAV gene therapy by the time they reach adulthood* How hiding a virus inside one of your own organelles gets it past the immune system* Why vaultAAVs don’t just dodge antibodies but also boost transduction up to 5x (in some cell lines)* Why gene therapies cost $850K to $3.5M a dose, and how synthetic biology could make them cheap* How a high-school science fair peptide led to Logan getting a minor planet named after him* How we can engineer adenoviruses to cross the blood-brain barrier in order to treat neurodegenerative disease and mental illnesses* What it means that our cells are full of a mysterious organelle nobody can fully explainGene editing technologists like David Liu, Feng Zhang and my friends in the AbuGoot lab have spent two decades proving that CRISPR-based systems can fix faulty genes. The much harder problem is how we can get gene therapies into everyone who needs them, and eventually make them as safe, cheap and ubiquitous as Ozempic.To do that, we need to be able to inject them more than once, and get them to our target tissues without the immune system getting in the way. That’s the problem Logan is working on.Watch on YouTube, listen on Apple Podcasts or Spotify.GUEST INFORMATION:* Logan Thrasher Collins, PhD, Washington University in St. Louis* Cathedral Therapeutics* VaultAAV preprint* Synthetic Biology for AAV Manufacturing* X (Twitter)CONNECT WITH US:* Website* Substack* YouTube* X (Twitter)* LinkedIn* TikTok Get full access to Frameshifts with Benjamin Arya at frameshifts.bio/subscribe

Most longevity talk is still stuck within two bad schools of thought. On one side, weak-sauce incrementalism: supplements, vibes and tiny effect sizes. On the other, immortality rhetoric that asks you to throw out the data altogether. Joe Betts-LaCroix has very little patience for either.He’s the co-founder and CEO of Retro Biosciences, one of the most ambitious longevity companies in the world. Launched with an initial $180 million of capital from Sam Altman’s personal wealth, the company has since raised new round led by 4P Capital at a $1.8 billion valuation. Retro’s mission is to add ten healthy years to human life by going after the cellular mechanisms of aging directly. Betts-LaCroix is an unlikely person to be running a longevity company. He left school with a D average and spent six years in a shared house with a “bunch of musicians and artists and weirdos”, doing electronics work to pay rent. That somehow became biophysics at Harvard, MIT and Caltech, then building the world’s smallest personal computer (OQO, acquired by Google), and later founding an automated animal-research company (Vium, acquired by Recursion), before landing on the question he couldn’t walk away from, which is why we age and what we can actually do about it.A lot of our conversation is about what a serious longevity biotech company looks like once you strip the noise away. We dig into how Retro chose its three core programs, autophagy, partial reprogramming and blood stem cell replacement, out of an initial six. We talk about how the company walked away from their blood plasma program despite promising early data, and how their liver reprogramming results, dramatic as they were in mice, became commercially unviable once GLP-1 drugs reshaped the entire liver-disease market. Their first drug, a small-molecule autophagy therapy aimed at neurodegeneration, has since begun dosing patients in a Phase I trial in my home country of Australia.Then we go deep on GPT-4b micro, the protein-design model Retro built with OpenAI. Joe lays out what it actually is and isn’t, how it can rewrite up to 80% of a protein’s amino acids while still preserving fold and function, and why the redesigned Yamanaka factors it produced have already made their way into FDA submissions. It’s striking to me that this model (among several others in its class) keeps proposing designs no human would ever have come up with, and that general-purpose protein generation carries a biosafety tension Joe doesn’t try to dodge.From there, our conversation opens up into the bigger questions: why aging may not need to be formally recognized as a disease for meaningful therapies to win approval, what it takes to build a company around a problem this hard, and, near the end, what happens to human ambition and our sense of time when people stop living only for tomorrow.In this episode, we cover:* Why the target isn’t immortality but ten additional healthy years* Why one of its first clinical programs targets autophagy in neurodegeneration* How partial reprogramming evolved from whole-body proof of concept toward tissue- and cell-type-specific intervention* Why Retro walked away from a promising liver reprogramming program despite dramatic mouse data* How GPT-4b micro works as a protein-design model, and why it’s more than “an LLM for Yamanaka factors”* Why protein design may unlock improved versions of our own endogenous maintenance machinery* How Joe thinks about regulation, aging endpoints, and what will eventually force the FDA to adapt* Why meaningful missions tend to produce stronger companies and stronger teams* Why he sees longevity not as a billionaire side quest, but as one of the most important things a civilization could work onThe most important problems in biology might also, if you’re careful about which ones you take on, turn out to be the solvable ones. GUEST INFORMATION:* Joe Betts-LaCroix, Co-Founder & CEO, Retro Biosciences* Retro Biosciences* OpenAI x Retro Biosciences, GPT-4b micro* Defining a longevity biotechnology company (Nature Aging)* X (Twitter): @bettslacroix* LinkedInCONNECT WITH US:* Website* Substack* YouTube* X (Twitter)* LinkedIn* TikTok Get full access to Frameshifts with Benjamin Arya at frameshifts.bio/subscribe

Plenty of people will tell you AI is about to design our drugs. Alex Zhavoronkov already has. His company’s lead candidate, rentosertib, is the first drug with both its biological target and its molecule discovered by generative AI to post real Phase 2 results in humans. That gap, between the pitch deck and a drug in an actual person, is the whole reason this conversation is worth your time.Alex started out in GPUs and high-performance computing, then left to bet that AI could design drugs better than people can. He retrained in biomedical science, and nearly two decades on, his company Insilico Medicine has gone public in a US$292 million Hong Kong IPO and signed a collaboration with Eli Lilly worth US$2.75 billion. That track record is what makes his bluntness about the rest of the field worth hearing.We get into the only benchmark he thinks matters: the time it takes to go from target identification to a real developmental candidate, and why he treats funding rounds, papers and patents as mostly noise. We talk about how Insilico initially stalled while chasing aging “in the most hardcore way possible”, why he’s convinced that experimental validation rather than model size is still the real bottleneck in biology, and the claim that tends to make founders in Boston and San Francisco uncomfortable: that if you’re not competing and collaborating in China, you’re already behind.In this episode, we also get into:* What “pharmaceutical superintelligence” actually means, and what it doesn’t* Why medicinal chemists aren’t getting replaced any time soon* Why humanoid robots are the wrong bet for the lab right now* Where embodied AI actually fits into biology* Why owning the best software doesn’t make a drug company defensible* How small, validated models end up training the bigger ones* Can you still compete in biotech without building in China?* When AI can design the drug, who actually captures the value?In Alex’s ideal version of all this, you wouldn’t even open a drug discovery platform. You’d just tell a language model what you want cured, for whom, and with what tradeoffs, and let it go to work behind the scenes. It sounds like science fiction, right up until it starts working.GUEST INFORMATION:* Alex Zhavoronkov, CEO Insilico Medicine* X (Twitter)* LinkedIn* Deep learning enables rapid identification of potent DDR1 kinase inhibitors (Nature Biotechnology, 2019)* Rentosertib Phase IIa results in idiopathic pulmonary fibrosis (Nature Medicine, 2025)CONNECT WITH US:* Website* Substack* YouTube* X (Twitter)* LinkedIn* TikTok Get full access to Frameshifts with Benjamin Arya at frameshifts.bio/subscribe

Many of science’s most important projects fall through the cracks between academia and industry. These are the foundational tools that could accelerate entire fields, and yet don’t fit neatly into a PhD thesis or a venture-backed startup.And to humanity’s credit, there has been growing interest in the last few years in building alternative systems for accelerating science beyond academia and industry.Groups like Episteme, the Arc Institute and Astera differ in structure and ambition, but they share a common premise: some of the most important work in science requires institutions that neither universities nor venture-backed companies are built to support.They also share something else. None of them would exist without billionaire philanthropy: Altman and Masa in the case of Episteme, Jed McCaleb in the case of Astera, and the Collison brothers in the case of the Arc Institute.And yet when I bring these organizations up with friends and colleagues in science, I often sense the same underlying skepticism: impressive as they are, can any of them really become durable drivers of scientific progress or are they structurally incapable of becoming anything more than donor-dependent experiments? Bell Labs changed the world, but it was built on an economic foundation, a telecommunications monopoly, that no longer exists. So, what’s left in humanity’s armament for progress? Well, there’s the NIH, DARPA and ARPA-H in America, ARIA in the UK, university-affiliated research institutes around the world, a dense ecosystem of startups concentrated in the major entrepreneurial hubs, and then a handful of billionaire-backed nonprofit research orgs. But there is another model that has been gaining traction in recent years: the Focused Research Organization, or FRO. These are nonprofit research organizations built around tightly scoped scientific milestones, typically with 10 to 30 person teams and budgets in the $20-30 million range.Adam Marblestone is the founder of Convergent Research and the architect of the FRO model. Late last year, in what many saw as validation of the model, the National Science Foundation announced a new initiative to “launch and scale a new generation of transformative independent research organizations to advance breakthrough science”.In my chat with Adam, he traces his path from graduate training in George Church’s lab to DeepMind’s neuroscience team. He came to believe that science needs a third institutional model, one that complements rather than replaces academia and industry.We discuss the idea of “intellectual dark matter”, the promising ideas researchers have but rarely get the chance to pursue. Adam explains why mathematicians need robust software infrastructure just as much as astronomers need telescopes, and how Convergent Research is systematically identifying more than 100 missing “Hubble Space Telescopes” across scientific fields. Adam argues that many breakthrough ideas remain invisible not because they are wrong, but because the shared infrastructure needed to test them does not yet exist.Topics we cover include:* Why progress in fields like mathematics and neuroscience is often bottlenecked by missing shared infrastructure (e.g. proof verification, connectome mapping, ultrasound brain interfaces)* How “intellectual dark matter” exposes systemic blind spots in the way science is funded, evaluated, and organized* How the Gap Map is systematically cataloging hundreds of missing foundational capabilities across scientific disciplines* Why building scientific infrastructure often requires industry-style execution inside nonprofit structures* Why some of the most ambitious deep-tech efforts are too infrastructural for venture capital, yet too operational for academiaGUEST INFORMATION:* Adam Marblestone, Founder & CEO of Convergent Research* Convergent Research* Gap Map* X (Twitter)CONNECT WITH US:* Website* Substack* YouTube* X (Twitter)* LinkedIn* TikTok Get full access to Frameshifts with Benjamin Arya at frameshifts.bio/subscribe

Parag Mallick is a Stanford Professor and Chief Scientist at Nautilus Biotechnology, a publicly traded biotech company. He’s also a professional magician and circus performer, which might sound random until you realize that his company does the closest thing to proteomics magic I’ve ever seen. The things that make you weird, he argues, are exactly what let you see data differently than anyone else.Here’s what caught me off guard: we can only measure ~8% of human proteins with current mass spectrometry tools. The other 92%—what Parag calls the “dark proteome”—is essentially invisible to us.Why is proteomics so much harder than genomics? Three reasons. First, protein concentration. DNA is basically uniform—roughly the same amount per cell. Proteins range from one copy to a billion copies in a single cell, and no analytical tool can handle that spread. Second, dynamics. Your genome is relatively static over your lifetime. Your proteome changes every second of every day. Third, there’s no equivalent of amplifying proteins like PCR can do. You get what you get.Nautilus is tackling this with single-molecule detection. The technical approach is fascinating: they use DNA origami nanoparticles. Picture a Ritz cracker with a flagpole sticking out the top. That flagpole has exactly one attachment point. They coat sample proteins with one half of a click chemistry reagent (methyltetrazine), the nanoparticle has the other half (trans-cyclooctene), and they bind together. One protein per nanoparticle. These then self-assemble onto a flow cell with 100-nanometer landing pads, creating a hyperdense array of billions of individual protein molecules.Now comes the really clever part: protein identification. Traditional proteomics tries to build one highly specific antibody per protein—an intractable problem when you’re dealing with millions of proteoforms. Nautilus does the opposite. They intentionally build cross-reactive affinity reagents. The system uses ~300 different affinity reagents, each one recognizing just a three-amino-acid epitope. That’s deliberately non-specific. Then they run 300 cycles of iterative mapping. The primary data looks like fluorescent NGS—you get a light-up at each location on the array or you don’t. Binary. Yes or no.The median protein only needs ~12 epitopes to be uniquely identified, but each protein gets touched 10-30 times across the 300 cycles for high confidence. It’s exactly like playing Guess Who: “Do you have glasses? Brown hair? A hat?” Each question alone tells you almost nothing, but together they pinpoint exactly who you are. Same principle here: “Do you have this 3-amino-acid sequence? What about this one?”The 300 binary measurements create a point in 300-dimensional space. Each protein has a characteristic signature in this space. The machine learning layer compares your observed pattern against the reference proteome and asks: what protein is compatible with this specific binding pattern? If you find a totally new protein, it’ll occupy a new point in that space—something not in the database.But here’s the thing about building something this audacious: you can’t prove it works before you start. Parag shared what the early days were actually like—renting a single lab bench at Stanford’s StartX incubator, trying to convince people to join when he couldn’t demonstrate single-molecule deposition yet, couldn’t show them they could run 180 cycles because they didn’t have an instrument. The first automation system was literally called “Parag” because he was pipetting by hand. How do you hire people to believe in something impossible? You share the vision of what it could mean—bringing the proteome to everyone—and see if that resonates. Some people thrive in that uncertainty. Others are brilliant at early-stage innovation but different people excel at scaling and productization. That evolution isn’t failure, Parag argues. It’s healthy. It’s part of the journey.We’re really just at the beginning of understanding biology. The genomics revolution, as transformative as it’s been, was the opening act. The era where we can actually see what proteins are doing—that’s what comes next.And in case you’re short on time, here’s a quick teaser:Watch on Youtube; listen on Apple Podcasts or Spotify.Guest Information* Parag Mallick. Professor at Stanford University & Chief Scientist at Nautilus Biotechnology.* LinkedIn* X (Twitter)* Nautilus Biotechnology* Large-scale single-molecule analysis of tau proteoforms* Proteomics Toolkit Paper (Nature, 2012)* High-density and scalable protein arrays for single-molecule proteomic studies (bioRxiv, 2022)Connect With Us* Website* Substack* YouTube* X (Twitter)* LinkedIn* TikTok Get full access to Frameshifts with Benjamin Arya at frameshifts.bio/subscribe

The future of human biology isn’t about better animal models - it’s about building human organs on chips that think like the body does.Don Ingber, founding director of Harvard’s Wyss Institute, has one of the most unconventional origin stories in the field. In 1975, as a Yale undergrad, he took a sculpture class where students built floating structures held together only by tension; no rods touching, no rigid frames. At the same time, he watched cells in culture flatten, round up, and then flatten again.The connection clicked: cells aren’t rigid blocks; they’re dynamic tensegrity structures, governed by forces, tension, and mechanical cues. That conceptual shift reshaped his view of biology.Fast-forward, and Ingber’s team would eventually build something that changed the definition of a “model system”: organ-on-chip devices the size of a thumb drive, lined with living human cells, that recapitulate organ-level physiology with astonishing fidelity.Key Takeaways:* Animal testing has a low success rate: 70% of clinical trials fail, with neurology hitting 95% failure rate. Organs-on-chip technology offers an alternative method that can help predict patient treatment responses* How a chip mimics an organ: Two microfluidic channels with different tissue types separated by a porous membrane. Cyclic suction mimics breathing or peristalsis* The microbiome breakthrough: Culturing complex bacterial communities for days using flow and mechanical forces creates a microbiome-like culture compared to static cultures* Personalized medicine economics: Test 100 patient-derived chips, identify the 50 responders, screen for toxicity, run focused clinical trials on the 35 most promising candidates* The Wyss origin story: A donor who walked away for two years, and a Martha’s Vineyard meeting that led to a $125 million gift* Why institutional structure matters: Independent governance, no deans, separate finances from Harvard and MIT* The Boston/Cambridge singularity: Universities, hospitals, VCs, pharma all in walking distance. You can’t recreate this by distributing talent globally* What’s at stake now: Immigration restrictions and funding cuts threaten the human capital that drives American innovationThis conversation goes beyond the science. We get into what it actually takes to build an institution like the Wyss, and how we’re watching the American scientific leadership face an existential threat with immigration restrictions and ideological constraints.And in case you’re short on time, here’s a quick teaser:Watch on Youtube; listen on Apple Podcasts or Spotify.GUEST INFORMATION:* Don Ingber - Founding Director, Wyss Institute for Biologically Inspired Engineering at Harvard University* “Cellular tensegrity: defining new rules of biological design that govern the cytoskeleton” - Journal of Cell Science (1993)* “Human organs-on-chips for disease modelling, drug development and personalized medicine” - Nature Review Genetics (2022)* X (Twitter) * Wyss Institute* Emulate (Organ Chip Company)CONNECT WITH US:* Website* Substack* YouTube* X (Twitter)* LinkedIn* TikTok Get full access to Frameshifts with Benjamin Arya at frameshifts.bio/subscribe

This week, I sat down with Shelby Newsad, Partner at Compound VC, to break down what makes a great biotech investment and why so many platform stories fall short.Shelby lays out a playbook forged across therapeutics, diagnostics, and lab automation. We dive into why mechanistic understanding predicts success, how diagnostics evolve into data businesses, and why the next wave of value may come from closed loop automated labs that combine hardware and software to finally link the whole bench.We explore faster paths to human evidence through compassionate use, phase 0 trials, and cassette microdosing, and how biobanks move discovery closer to patients. Shelby also explains how RNA secondary structure targeting with covalent small molecules could merge phenotypic and target based discovery inside one company.Then things get wonderfully weird. We discuss manufacturing as value capture in a world where intelligence designs drugs, gene edited flowers as a low regulation high margin venture canvas, and what it means when biohackers and LLMs push healthcare outside traditional borders.Key Takeaways* Mechanism > platform: Companies prosecuting mechanistic disease biology have a 3–4× higher chance of success than platform first bets with fuzzy “why.”* Diagnostics as engines: Tests that are profitable and build biobanks create data and network effects and decouple cost from revenue via services and licensing.* Human evidence faster: Use compassionate use, phase 0, and cassette microdosing to de risk earlier with real PK PD and binding data.* Automation gap: Most lab robots lack vision systems. Closed loop labs that are miniaturized and observable are the usability unlock.* RNA covalency: Covalent binders to RNA secondary structures offer clean transcriptome readouts blending phenotypic and target based discovery.* Value capture shift: If AI designs the drug, manufacturing of proteins, chemicals, and gene therapies could own the margin stack.* Global arbitrage: The United States leads, but Australia and China accelerate timelines. Trials and approvals follow speed and clarity.* Weird is good: Gene edited flowers that are non edible and simpler to regulate marry beauty, margins, and scale, unexpected but venture real.But in case you’re short on time, here’s a quick teaser:Watch on YouTube; listen on Apple Podcasts or Spotify.🔔 SUBSCRIBE so you don’t miss our next conversation!GUEST INFORMATION:* Website* LinkedIn * X (Twitter)CONNECT WITH US :* Website* Substack* YouTube* X (Twitter)* LinkedIn* TikTok Get full access to Frameshifts with Benjamin Arya at frameshifts.bio/subscribe

What happens when the man who launched commercial spaceflight decides humanity needs to stop aging?Peter Diamandis is building the infrastructure to tackle aging. He founded the XPRIZE Foundation, created Singularity University with Larry Page and Ray Kurzweil, and wrote the playbook on exponential thinking with Abundance and Bold. However, somewhere between flying Stephen Hawking into zero gravity and recruiting Elon Musk to his board, he realized something: if we’re building an unlimited future, we need to be around to see it.We talk about why Peter walked away from space (his first love) to focus on longevity. He shares his story on how he raised $101 million for the XPRIZE Healthspan, the largest incentive prize he’s ever launched, and why 730 teams are now racing to reverse functional aging by 2030. Peter also talks about what it means to restore cognition, immune function, and muscle capacity to what you had 20 years ago, measurably and reproducibly.We also get into the economics of abundance where Peter states that artificial intelligence, the most powerful technology in human history, is becoming free for 8 billion people. Intelligence as a service is demonetizing at 79% per year. Ultimately, are we heading toward Star Trek (exploring the universe with godlike capabilities) or WALL-E (sitting back while robots feed us grapes)?Key Takeaways:* XPRIZE Healthspan: $101M prize with three targets: reverse cognitive decline, immune exhaustion, and sarcopenia. Winner announced by 2030.* The six D’s of exponentials: Digitize → Dematerialize → Demonetize → Democratize → Deceptive → Disruptive. Every industry that touches ones and zeros follows this curve.* Why biotech is broken: Wall Street stopped valuing potential and started demanding revenue. Peter thinks high-fidelity AI cell models (the biology equivalent of physics-based simulations of SpaceX’s Falcon 9) could restore confidence before Phase 3 trials.* Healthcare will be free: AI diagnosticians already outperform human doctors (92% vs 72% accuracy). Data collection will be ambient: sensors in your watch, ring, breath, typing patterns. Companies will pay to keep you healthy because catching disease early is cheaper than treating it late.* The cost curves: Full-body MRI machines that don’t need helium cooling, cost $200K instead of $2M, and scan in 10 minutes instead of 90. Genome sequencing headed toward free. Grail’s multi-cancer detection becoming a volume play.* Why Moore’s Law mattered more than anyone realized: In 1958, Gordon Moore noticed transistors per dollar doubled every 12-18 months. That gave us 60+ years of predictable exponential compute. Everything else, including AI, biotech, and nanotech, runs on top of that curve.* The transition from space to longevity: Peter realized rockets and satellites were incremental, but if AI, 3D printing, and nanotech go exponential, space becomes easy downstream. So he went deep into the enabling technologies first.* What “solving everything” looks like: Enough compute, enough algorithms, enough data to uncover the secrets of math, physics, chemistry, biology. Imagine what happens when every day you wake up to a scientific breakthrough?This is one of those conversations where you realize the future isn’t decades away, it’s deceptively close. The tools exist, the compute exists and intelligence is free. The only question is whether you’ll build something with it, or watch from the sidelines.And in case you’re short on time, here’s a quick teaser:Watch on Youtube; listen on Apple Podcasts or Spotify.GUEST INFORMATION:* Peter Diamandis - Founder of XPRIZE, Singularity University, Bold Capital Partners* Book: Abundance, Bold, The Future Is Faster Than You Think* Upcoming Books: We Are As Gods: A Survival Guide in the Age of Abundance* X (Twitter)* Moonshots Podcast: Weekly discussions on exponential technologiesCONNECT WITH US:* Website* Substack* YouTube* X (Twitter)* LinkedIn* TikTok Get full access to Frameshifts with Benjamin Arya at frameshifts.bio/subscribe

What happens when one of the world’s leading neurosurgeons steps away from the operating room to become a first-time founder? Dr. Theodore Schwartz pioneered minimally invasive brain surgery going through the nose and eyelid instead of opening the skull. He’s published over 200 papers, written a bestselling book on neurosurgery history, and now he’s building a MedTech startup to solve a problem that’s plagued him for decades: there’s no way to continuously monitor a patient’s pupils during surgery.This conversation goes deep. We talk about the psychological weight of knowing you might paralyze 3-4% of patients no matter how good you are. Why it took him 20 years to truly master neurosurgery. The physics problems that keep him up at night. And why he thinks AI will struggle with surgery for a long time, not because of intelligence, but because of how differently humans and computers evolved.We also get into the future of brain-computer interfaces. Ted sits on the scientific advisory board of Precision Neuroscience, and he breaks down exactly how their approach differs from Neuralink and Synchron. Spoiler: it’s about electrodes on the surface of the brain, not inside it, and why that might actually be better for getting high-bandwidth data out.Plus: focused ultrasound opening the blood-brain barrier, targeted immunotherapies for glioblastoma, the doorman who changed his perspective on hard work, and what it’s really like to transition from 40 years of clinical practice to writing FDA guidelines and pitching to VCs.* Why continuous pupillary monitoring through closed eyelids could revolutionize neurosurgery and ICU care* The three competing BCI approaches: Synchron (in blood vessels), Neuralink (penetrating electrodes), Precision (surface electrodes)* Why neurosurgery took 25 years for even the best hospitals to adopt transorbital approaches* The decision algorithm surgeons use when removing a tumor might blind someone, and why AI can’t make that call yet.* How Harvey Cushing reduced brain surgery mortality from 50% to 8% and essentially founded the field* Why grit and curiosity matter more than raw intelligence in neurosurgery* The transition from practicing clinician to founder: patents, FDA meetings, grant writing, and learning to create value before raising VC money* Why infrared light passes through eyelids better than visible light (it’s the melanin)How focused ultrasound might eliminate the need for opening skulls entirelyThis is one of those conversations where you realize how much frontier work is still happening in medicine, and how the people pushing those boundaries think about risk, responsibility, and what it means to literally hold someone’s brain in your hands.Watch on Youtube; listen on Apple Podcasts or Spotify.GUEST INFORMATION:* Theodore Schwartz, MD - Neurosurgeon, Author, Entrepreneur* Book: Gray Matters: A Biography of Brain Surgery* Website* X (Twitter)CONNECT WITH US:* Website* Substack* Youtube* X (Twitter)* LinkedIn* Tik Tok Get full access to Frameshifts with Benjamin Arya at frameshifts.bio/subscribe