About
As co-Founder and Partner at NextView Ventures, I take a high conviction hands-on…
Articles by David
Activity
7K followers
-
David Beisel shared thisEvery AI tool being sold into venture right now is a running coach. It writes a sharper plan for an athlete whose ceiling is fixed. A better sourcing mechanism, a cleaner first draft of the investment memo, a "non-voting AI IC" seat at the Monday meeting. The plan around the engine gets sharper. Yet the same old VC engine stays exactly what it was. I personally have run the same half marathon repeatedly for a decade, chasing a time that I've never been able to hit. The last two seasons I handed my training over to an AI coach app. Admittedly, the plans were genuinely better than anything I wrote for myself. But I still didn't beat my old number because the constraint was never the plan. It was me! That is the trap inside "AI makes our VC partners more productive." The gains are real, and yet they are incremental, because the athlete is always the ceiling. Running is a domain where the human is genuinely the point. Venture only assumes it is.
-
David Beisel shared thisAI disruption is demanding every venture capital firm to do what Radiohead did in 1999: put down the instrument that made them. Same musicians. New instruments. Most of the industry is getting the shape of AI change wrong. Radiohead had just made the album OK Computer, the record that made them famous for guitars, and the obvious next move was another one just like it. Instead they made Kid A. Jonny Greenwood traded lead guitar for synthesizers. Ed O'Brien turned his into a machine of loops and effects. O'Brien said it plainly: "It's scary. I'm a guitarist and suddenly there are no guitars on this track." The easy version of an AI-enabled VC firm is a new paint job: keep playing guitar, add a synth in the back of the mix, call yourself AI native. That was O'Brien's instinct, and it loses for the reason his did. You get a slightly better version of the record you already know how to make. The real version is harder. The musician is durable; the instrument is disposable. The VC musicianship is the judgment, the eye for a company worth backing. The instrument is whatever you do all day to apply it. Change instruments and you're still a musician. I've been living this one myself. The past four months I've spent inside Claude Code, building workflows that do by machine what I used to do by hand. A year ago that was zero percent of the job. The judgment carried across intact. The daily instrument did not. The hardest resistance might not come from the partners doing the rebuilding. Radiohead's own label was alarmed, shipped Kid A with no singles and no videos, and called it a "business challenge." The album debuted at No. 1. The people whose job was to sell the last record couldn't hear how the next one would sell. --> The VC firms that come through this transformation won't be the ones that hire AI specialists to play alongside the partners. They'll be the ones where the partners themselves put the guitar down. <-- Read my newest Carried Away blog post in full, No Guitars on this Track, linked in the comments below.
-
David Beisel shared thisOne of the most beautiful things I saw in fifteen hundred miles of cycling was a warning. My current pair of Carried Away posts carries this warning for VCs. The medieval walls of York, England stand on buried Viking earth, which stands on buried Roman stone. Nobody ever started over. For centuries, building on top of the previous layer, again and again, was exactly the right method. Then the premise underneath them died, and no new layer could save it. The walls survived by becoming a museum of themselves. Venture firms are building York walls right now. Ask a firm what it's doing about AI and you'll hear about the newest layer: the sourcing tool feeding the same old funnel, the model that pre-screens what associates used to screen. Each course laid carefully on the inherited structure, because building on top costs less than asking whether the premise underneath still holds. That buried premise is that partner judgment was scarce and human. The warm intro, the screening pyramid, the Monday meeting all exist to ration access to it. Scarcity is the thing AI ends. So the test of any firm's AI initiative is simple: does the rationing structure survive it? If it does, it was just another layer, however it was pitched. York got two thousand years out of building on top. A venture firm will not get twenty. So make the move the walls never made. Don't automate the firm you have. Build the one you'd build today from scratch. Treat a fully AI-run firm as the default, and add people back only where the machine visibly fails without them. Then leave the old firm standing, the way York left its walls, and walk your proprietary judgment and your brand out the gates onto open ground. The firms that come through this transition will be the ones that could tell what was load-bearing from what was only old. The full argument runs across two blog posts, both out today in the comments below.
-
David Beisel shared thisAndre Retterath's new Data Driven VC Landscape report, genuinely the best map we have of AI adoption in our business, has the most honest finding I've read: asked what's holding VC firms back, they didn't name the tools they're missing. They named time. Then it prescribes a tool. And so do the rest of us. The constraint on becoming an AI-driven firm is the attention of the senior people who would have to sit down and teach a system what the firm believes. That part isn't for sale. You can buy the software, license the data, hire the engineer. None of it arrives knowing what your partnership counts as an exceptional company, or when to step into a struggling portfolio company. That's the firm's taste, and only the people who hold it can install it. We keep budgeting as if the fix were on a price list, because that's what management fees buy. AI doesn't remove venture's bottleneck. It relocates it. The scarce partner-hours that used to go into making the calls one at a time now go into encoding those calls once, so the machine can make the routine ones at scale. Some firms will spend that attention now, deliberately, at the cost of deals they could have chased. Everyone else will keep adding line items and waiting for the tools to install themselves. My new Carried Away post, on the one thing an AI-driven firm can't put on the invoice: https://lnkd.in/dJdzua2K Link to the Data Driven VC Landscape 2026 report: https://lnkd.in/dETUscvG
-
David Beisel shared thisEvery venture firm keeps a private list of the great companies it passed on. The more painful list is the one no one keeps: the companies it never saw at all. The first list was written by judgment. The second was written by the calendar. The scarce resource in any venture firm is a senior partner's time, and a single miss can cost the whole fund. So the industry buys coverage: pyramids of associates screening upward, accelerators standardizing intake, scout networks writing checks beyond the partnership's reach. Coverage has always been for sale. But every one of those structures is a copy of partner judgment, and every copy is lossy. The accelerator trades resolution for volume. The scout carries a copy out of the building. The junior investor copies it by apprenticeship, the slowest and lossiest copy of all. Venture has never had a way to scale judgment except by copying it, and every copy was lossy. That is the copy problem. What AI changes is the copy. Codified judgment is digital, and it duplicates at constant fidelity. Company number two thousand gets screened by the same judgment as company number twenty. No apprenticeship ever could. Solve the copy problem and the bottleneck moves back to where everyone always claimed it lived: judgment itself.
-
David Beisel shared thisEvery "AI plus the human partner" setup in venture capital assumes the same thing: that the partner stays essential for years. Chess made that assumption too. It held for twenty years. In Go it held for less than two. The "centaur era" is the stretch when a human paired with a machine beats either one alone. Dario Amodei used the term recently for software engineers, and he is right that we are in one. He is also right that it is temporary. The real question is what decides how long it lasts. The answer is legibility. Where a person can spell out and check the machine's work, the era runs long: chess, supervised driving. Where the best judgment is intuition no one can fully articulate, the machine crosses the threshold and there is often little left for the human to add. That was Go. It lasted barely two years. In venture, the part of the job that sets the returns is the illegible kind: the read on a founder, the call on a market that does not exist yet. Two good partners study the same founder, disagree, and cannot settle it by trading reasons. That is the tell. Once a model gets there, there is not much legible left for the partner to add on top. We do not actually know we are the chess case, and we have reason to suspect we are the Go case. The centaur era of chess lasted twenty years. The centaur era of Go lasted nearly zero. In VC, we do not get to choose.
-
David Beisel shared thisThe risk isn't where you live on the valuation curve. It's living there by accident. That's the uncomfortable gut check in Rob Go's piece, written right after we at NextView Ventures leaned into a $300M seed at the very top of the curve. You can win at the top or the bottom, but only if you actually *chose* where you're standing. It's easy to drift a full quartile (or more!) over a decade and never notice your model moved you there. Naming the bet out loud is the discipline.David Beisel shared thisIt seems timely to write this post after we just invested in a $300M seed at a rich valuation: Where Do You Live on the Valuation Curve? Carta released an interesting graph showing seed valuations by percentile over time. The TLDR - everything is going up, but the spread between the top percentiles and the lower percentiles has increased dramatically. Every investor tends to live somewhere on this curve. But as the curve has shifted, where an investor lives may have shifted quite significantly over time. One may find themselves pretty much only looking at companies in the bottom 25% valuation range because that's what your model is tuned to. But perhaps 10 years ago, you were mostly looking at companies in the top 25th% percentile range. Sure, the prices you paid went up, but probably not nearly as much as the market has moved. Is that good or bad? You can win either way, but you need to know where you live and what kind of bet you are making as a result. Full post in the first comment.
-
David Beisel shared thisWalden Robotics launches today with a $300M seed round at a $1.1B valuation, and NextView Ventures is proud to have leaned in with a meaningful check. Here's why we had the conviction ⬇️ It started with Dave Johnson, a founder we backed once already as early investors in Dexai Robotics before his exit in 2024. Dave could write his own ticket, and instead of starting another company he's joining Walden as Chief Product Officer. When a founder of that caliber builds inside someone else's company, that team has to be world class. That team is led by co-founder and CEO Russ Tedrake: Toyota Professor at MIT, creator of the Drake open-source framework, and most recently SVP of Large Behavior Models at Toyota Research Institute. The core idea: instead of being programmed by hand, these robots learn new physical skills from human demonstration, the same leap LLMs made for text. Russ's team showed pretrained models learn new tasks with 3 to 5x less data, and the approach already drove Boston Dynamics' Atlas through multi-step jobs with no new code. Walden is spinning that work out of TRI. My partner Lee and I spent real time at their Cambridge HQ watching the robots work, and seeing it in person turned interest into conviction. They're already deployed alongside people at one of the world's largest manufacturing plants, going from first pilot to full deployment in under two months. The round is co-led by Toyota and Deviation Capital, with NVIDIA and Samsung as strategic partners. It also brought a familiar face: Deviation's Colin Beirne, a longtime collaborator from our WHOOP co-investing days. The last few years of AI were about knowledge work. The next few are about the physical world. Excited to back Russ, Dave, and the whole Walden team.
-
David Beisel shared thisResistance to change is not an information problem. It is a wiring problem. Status-quo bias, loss aversion, present bias: evolution built them in, and knowing their names does nothing to loosen their grip. I knew all of this cold in 2001, watching the dot-com crash take the company I had helped build, and I still did not take it well. Here is what unsettles me now. Venture capital's whole job is capitalizing on this wiring in other people. Every pitch memo we write is a bet that the incumbent's brain won't let it adapt before the startup grows too big to stop. We are, professionally, connoisseurs of everyone else's wiring. Then AI walks into our own maze. The same brain that spots disruption coming for every other industry goes quiet when the disruption is coming for us. The same organ returns opposite verdicts, and the only thing that changed is whose business is on the table. Mine included. I can name every bias in the room and still struggle to loosen their grip. New post in Carried Away, my Substack on how AI is upending venture capital: The Same Brain. Link in the first comment. Tell me where I'm fooling myself.
David Beisel replied to a comment
10h
Nickhil Bhave, your Gong point rightly implies what replaces the human can't be sold to the human. That's why Gong ships a coach, and why every AI tool in venture is a coach, too. The buyer is the athlete. Which is why the substitution rarely comes from the incumbent. It most often comes from whoever sells to the person one level up who actually owns the outcome and never ran the miles.
David Beisel commented on a post
16h
The reframe for me as a layman: the hard part isn't only choosing the therapeutic gene, it's designing the control system around it. Narrowing thousands of promoter candidates in software before anything hits the bench is exactly where this kind of model should earn its keep.
-
David Beisel liked thisDavid Beisel liked thisA few weeks ago, a "cool” AI use case" went around: connect it to your family's calendars, and have it generate a podcast for daily school drive, narrating your kid's soccer game, their birthday, whatever else it's decided you should know. Today, NextView Ventures is announcing our investment in Fambot, an AI chief of staff that reads the emails, calendars, and group chats flooding a parent's life and turns them into a clear plan for the day. I’m certainly not alone in saying I don't want to hand off car chats with my kids to AI. I want help decoding what matters from the barrage of teacher updates, PTA notices, birthday party invites, back-to-school nights, sports sign-ups, and WhatsApp group chats flooding my inbox and my brain. I want the headspace to ask my kids about their day myself, and really listen to their answers. For years, an accessible solution to help parents offload the overhead and be more present felt like magical thinking. But as AI became capable of handling real tasks, so did the opportunity to help tens of millions of busy parents at scale. The best consumer AI companies will abstract the algorithms and mental overhead we run silently, freeing us to stay in the moment and be more human. As a parent, that overhead is the running list of to-dos that stand between me and the parts of parenting I don’t want to outsource: quality time with my kids. That's what I love about Fambot. ❤️ 🤖 Fambot’s led by David Reich, a second-time founder, former Uber exec, and a dad of three living this exact chaos daily. He’s designed Fambot around an important recognition: parenting is rarely a single-player job. In many households, one parent becomes the default admin while the other ends up genuinely on the outside. Fambot runs quietly across everyone's calendars and inboxes so no matter how the workload gets split, everyone stays in the loop. Fambot launches publicly today, just as families across the country head back to school. More than 1,000 families (including mine!) used it during the private beta, and now it’s live on iOS and Android. NextView is proud to have co-led Fambot's $3.5M pre-seed round at inception, alongside Kate McAndrew and Baukunst, and joined by Correlation Ventures, Karman Ventures, and Founders Network. Congratulations to Dave, Greg Karlin, Jason Morrow, and the whole Fambot team. Link in the comments to read more about NextView’s investment. 👇
-
David Beisel liked thisDavid Beisel liked thisI've been at Vanta for just under two months now. I've caught up with a lot of friends and colleagues and everyone asks, "How is it?". My response has always been the same, "Intense, in the most wonderful way." It's moving fast among thoughtful, talented people with incredible momentum behind us and a lot of work ahead of us. How we shape our work here can change how technology is built globally. I'm hiring multiple roles for my team. If you know anyone or are interested yourself, please let me know. Links are in the first comment.
-
David Beisel liked thisDavid Beisel liked thisGene therapy isn’t just about choosing the right therapeutic gene. It’s also about controlling where, when, and how much that gene is expressed. At the #ReAgent2026 bio + AI hackathon, we asked a practical question: Can a model translate a gene-therapy requirement into promoter sequences ready for experimental evaluation? We built the prototype during the hackathon and then turned it into a product. Introducing Promoter Atlas by Genomic Intelligence. Researchers specify: → Where a therapeutic gene should be active → Where it must remain silent → The desired expression range Our models evaluate natural and designed promoters, returning prioritized sequences ready for synthesis and laboratory validation. We’re moving the broad search into software so experiments can focus on the strongest candidates. Read the full story, subscribe, and explore Promoter Atlas to compare candidates, or contact us about a private design program: https://lnkd.in/gPBmaGVk More updates are coming! #GeneTherapy #Biotechnology #AIForBiology #Genomics #DrugDiscoveryDesigning the Control Layer of Gene Therapy: Introducing Promoter AtlasDesigning the Control Layer of Gene Therapy: Introducing Promoter Atlas
-
David Beisel liked thisEvery AI tool being sold into venture right now is a running coach. It writes a sharper plan for an athlete whose ceiling is fixed. A better sourcing mechanism, a cleaner first draft of the investment memo, a "non-voting AI IC" seat at the Monday meeting. The plan around the engine gets sharper. Yet the same old VC engine stays exactly what it was. I personally have run the same half marathon repeatedly for a decade, chasing a time that I've never been able to hit. The last two seasons I handed my training over to an AI coach app. Admittedly, the plans were genuinely better than anything I wrote for myself. But I still didn't beat my old number because the constraint was never the plan. It was me! That is the trap inside "AI makes our VC partners more productive." The gains are real, and yet they are incremental, because the athlete is always the ceiling. Running is a domain where the human is genuinely the point. Venture only assumes it is.
-
David Beisel liked thisDavid Beisel liked thisOne of my closest friends from business school, Paraag Marathe, just left the San Francisco 49ers after 25 years. Paraag grew up in Saratoga, California. But his parents weren't engineers at Apple or IBM like many of the Asian immigrants who lived in Silicon Valley back then. He started working at his parents Round Table Pizza store when he was in middle school. He learned how to manage customers and talk to people from a very early age because he had to. He would go on to study business at UC Berkeley before becoming a management consultant at Bain & Company. While he was at Bain, he did an externship with the 49ers. That externship turned into a full-time role with the organization. He got his MBA from Stanford with me, while he continued to work for the 49ers nearby. Paraag changed the NFL. He made a game that was based on mysterious intuition and instinct, more analytical and statistics driven. Michael Lewis credits Paraag in Moneyball. Billy Beane from the Oakland Athletics often spoke with him. The legendary Bill Walsh trusted him with drafts and trades. He became respected around the league for his wizardry around analytics, but also across the negotiation table. From Albert Breer in Sports Illustrated: A quarter century later, Marathe’s mark on the organization has proven indelible. He survived good times and bad, going from working with Steve Mariucci to Dennis Erickson to Mike Nolan, Mike Singletary, Jim Harbaugh, Jim Tomsula, Chip Kelly and Shanahan. And maybe the most interesting thing, to me anyway, is how he’d be able to win over dyed-in-the-wool football guys who had every reason to stereotype or distrust Marathe as a numbers-cruncher who seemed to outlast everyone else in the organization. “Paraag is one of the smartest people I've been around,” [Kyle] Shanahan said Sunday via text. “You don't run into people in the NFL who have been in one place for 25 years. Paraag was here for 25 years—that's unheard of and speaks to how good he is at what he does. He's a very good friend of mine, one I am grateful to the 49ers for connecting me with, and one I hope is in my life forever.” “Paraag is just a force for the good,” Jim Harbaugh told me on Saturday. “He’s like an operator for the good, always cutting edge. … He's a connector. And a great teammate. You just always felt like Paraag had your back. … I mean, just so many examples of him being a resource for the good.” He did all of this while being one of the only people who looked like him across all of the front offices in the National Football League (NFL). Paraag was one of the few if not the only Asian American in a player operations executive role. Fans, the media and people around the league saw his face and assumed he was just a number cruncher. But he was far more than that, and everyone who actually interacted with him knew it. He is beloved and respected around the league for not only his talent, but who he is as a person. I can't wait to see what he does next.
Experience & Education
Volunteer Experience
Other similar profiles
Explore more posts
-
Jeff Immelt
Dayna Grayson is a leading VC who has “sat in every chair” at a young age. She founded Construct Capital with Rachel Holt, an early leader at Uber. Together, they are tackling the digitization of industry. Dayna will discuss macro trends, what it takes to be a founder, and the potential for AI. She is one of the few who have successfully blended a career of operations and investing. Check out our discussion at: https://lnkd.in/eYBUGjbW #leadershipdevelopment
9 Comments -
Daniel Dart
🚨NEW EPISODE: Recorded live at FUTURE TITANS 2026 - Jeff Perry of Carta sat down with the iconic Seth Levine, co-founder of Foundry. Seth has been in venture for 25 years, built Foundry from scratch as an emerging manager himself, and has backed about 50 emerging manager funds through his fund of funds. He has genuinely seen every side of this table. They went deep on building Foundry, why VCs are in the influence business, not the decision business, and why the concentration problem in venture is not only bad for LPs, but also for the innovation ecosystem overall. And why Seth's new book, Capital Evolution, is so important for the future of America. 🎧 Links to listen... Apple: https://lnkd.in/ehQUQ2EM Spotify: https://lnkd.in/eU4FExpg
1 Comment
Explore top content on LinkedIn
Find curated posts and insights for relevant topics all in one place.
View top contentOthers named David Beisel
14 others named David Beisel are on LinkedIn
See others named David Beisel