About
Swell Partners is a mission-oriented, growth-minded venture capital firm based in New…
Activity
6K followers
-
Jay Patil shared thisThe model is the magnetJay Patil shared this"The team you build is the company you build," Vinod Khosla and Keith Rabois have been saying it for years. When Jay and I first heard it, it was an oh shit moment because two of the best investors of all time were obsessing over the same thing we'd been obsessing over. You didn't hear that from VCs obsessing over memes, market maps, and thought pieces. You heard it from the ones who had spent decades building teams, seen it validated in the outcomes of their best companies, and made it a priority at the fund level. Every VC fund has a thesis, and Swell's is to invest at the intersection of AI & critical infrastructure. We've seen this play out in Fund I with Loft (Space & Defense) and Crisp (Supply chain), and Fund II is built entirely around it. But anyone can have a thesis on what other people do. Without elite founders and teams, a thesis is not a strategy, it is a blog post. At Swell we run a concentrated portfolio. Every company has to matter, every position has to be capable of returning the fund. There's nowhere to hide, which is the point. When you back that few companies, the work that happens between the check and the outcome is everything. For us, that work has always centered on hiring. Jay and I have placed more revenue hires across our portfolio than we can count. Not intros to recruiters, not candidates from a database. We sit in on interviews, we evaluate candidates against a framework we built from watching which hires actually built revenue from zero. That work has also made us better investors. The diligence gets sharper because the post-investment work is real. We evaluate founders from the outside, the team, the market, the timing, and from the inside, a decade of sitting across the table from every revenue hire and learning what actually predicts success. When you have both perspectives, the picks get better. That framework is now a product called Archetype, which takes a decade of hands-on recruiting and makes it always-on across the portfolio. It does three things at the fund level. It helps us win allocation because founders choosing between comparable seed checks pick the one that comes with a real edge in building the team. It helps us support companies through the gap between inception and Series A, where the solution is almost always the people. And it compounds, because every great hire we place becomes a referral source to more elite builders, a future founder, or a reference that strengthens the next search. At every level the core skill is the same: evaluating people and sharpening our judgment to be right and to be early. Most funds talk about platform and value-add (most of which is BS). We are building something that actually compounds. We didn't design it this way. We just kept doing the work and noticed the flywheel was already spinning. Concentrated portfolio, operational firepower, better pattern recognition, better picks, better hires, founders who win, founders who refer. The model is the magnet.
-
Jay Patil reposted thisJay Patil reposted thisWe just released the new version of Archetype, Swell VC's hiring evaluation platform for revenue roles built on a framework we developed over 10,000+ hours of founder and operator interviews, across a decade of seed investing and recruiting. The founder response has been stronger than we expected. Archetype now runs five dedicated scoring engines: Sales, Customer Success, Marketing, Business Development, and Operations. Each one is independently weighted and calibrated, because the traits that predict a great sales director have almost nothing in common with what makes a great ops leader. We're not expanding this to engineers, because revenue is what we know inside and out from years of recruiting across our portfolio. It also now evaluates candidates across multiple interview rounds, building a cumulative body of evidence, generating custom interview scripts that target gaps from prior rounds, and re-evaluating everything from scratch each time. By the final round you're not asking broad questions, you're pressure-testing the last remaining unknowns. You can compare candidates side by side with charts, and any scorecard can be shared and viewed with a link, no login required. The whole system is built so a founder can hire the best person in the country for the role, from IC to Director to C-level, and have the evidence to know it before making the offer. Every founder we back gets full access, we run it on every port co recruiting search, and it's become a proprietary asset of the fund deployed across the entire portfolio. Why does this matter? Because the right hire changes everything. "You win with people" has been the foundation since the beginning. We're opening it up to everyone for a limited window. Check it out, link below.
-
Jay Patil reposted thisWhy the most exciting opportunity right now is in the world’s most boring industry: The insurance industry wrote >$8 trillion in premiums last year (forecast to reach $9 trillion this year). Yet customer experiences remain poor, transaction and data processing are inefficient, and underwriting models still rely on rudimentary statistics. Right now, two major shifts are underway. First, the economy is generating entirely new categories of risk, from AI / autonomous systems to climate and advanced energy infrastructure. Second, tools that already exist, like predictive modeling algorithms and AI, are mature enough to be applied to insurance. It’s a rare moment where an industry this large has such clear room for transformation... and the tools to make it happen.
-
Jay Patil shared thisWe have turned our obsession into an actionable product - You Win With PeopleJay Patil shared thisWe built a sales assessment hiring tool at Swell VC called Archetype. It scores candidates against a proprietary framework we developed over 10,000+ hours of founder and sales interviews across 10 years of early stage investing, refined by watching which hires actually built revenue from zero and which ones flamed out across our portfolio. This week we shipped it as an API and MCP server and listed it on Smithery, which makes it the first sales hiring assessment available natively inside AI agents like Claude. This means any AI workflow or agent that touches hiring can now call Archetype like a native tool and score a sales candidate in seconds. A seed fund with two partners just shipped an MCP server before every HR tech company and VC on the planet. We didn't set out to beat them all to the MCP ecosystem, but here we are! We've shipped agents and sub-agents internally at Swell to run a growing fund without growing headcount, but this is the first external product we've built. When you're making your first or second sales hire at a seed stage company, getting it wrong costs you a year. The team determines everything. Archetype gives founders a way to pressure test candidates against patterns we've seen play out hundreds of times, the traits that separate closers from talkers, builders from those along for the ride. Anyone can try it at archetype.swell.vc. The MCP server is at https://lnkd.in/eMu_3S3p. Swell portfolio companies get full access free, forever, as part of what we do for our founders.
-
Jay Patil shared thisRecognizing greatness never gets old, pumped and honored to continue building Swell VC for the next decade and beyondJay Patil shared thisWe just launched a new site that embodies what we've been building: swell.vc Everyone's chasing AI. A lot of it doesn't matter. We invest where AI meets the stuff that does: enterprise, space & defense, bio, industrial, and commerce. The infrastructure the world actually runs on. Jay Patil and I have been at this since we joined an AI startup over a decade ago, when AI was still known as ML & NLP. We were a little early then. But the past five years, and looking forward, we've been right on time. Loft Orbital, Crisp, Deepnight, ScienceIO, IronGrid, Parametric, RyboDyn, Credal, and so many others. Founders building things that are hard to build and harder to replicate. We've placed 75+ key hires across the portfolio along the way. Turns out the right people create even more leverage in the AI age. Taste, empathy, agency, and obsession compound when the tools are this powerful. Come take a look, link in comments.
-
Jay Patil reposted thisJay Patil reposted thisThere are two philosophies in venture capital portfolio construction. One says: “We don’t pick winners. We pick everyone.” Write 50+ checks, pray one hits, collect fees along the way. Index the market, and claim to be good at picking when one of the lottery tickets hits. The other says: “We have the judgment to pick winners.” Ten to twenty companies. Every one has to matter. Every position has to be capable of returning the fund. There’s nowhere to hide. Most of the industry chose the first path. It’s safer for the GP. You spread risk, you diversify away your mistakes, and the management fees on a big fund pay well regardless of performance. You’re not optimizing for returns. You’re optimizing for AUM. We chose the second path. Venture capital is splitting in two: multi-billion dollar platforms that look more like asset managers, and small concentrated funds built on conviction. Last year, 65% of all VC capital flowed to just 30 firms. The data consistently shows that small, focused funds outperform mega-funds on returns. The math is structural: a $30M fund can generate outsized multiples from outcomes that wouldn't move the needle for a billion-dollar fund. 1 unicorn, 1 approaching, from a 12 company portfolio. Even the best-performing venture funds in the world hit unicorn rates in the single digits. We’re already there from Swell Fund I, with another one knocking on the door. When it crosses, two unicorns from twelve companies might be some kind of record for a fund our size. A lot still has to go right, and we’ve had some luck along the way, but we like where we stand. You can hide weak judgment inside fifty tiny checks. You can’t fake strong judgment inside twelve. When we back a founder, we’re betting the fund on them. They know it. We show up differently because of it. Concentration isn’t the risky strategy. It’s the honest one that drives returns.
-
Jay Patil reposted thisJay Patil reposted thisDetect. Classify. Remember. Track. Sovereign: Multi-domain Automated Targeting AI
-
Jay Patil posted thisSomething I’m learning about infrastructure investing: narratives don’t move budgets. In manufacturing, energy, defense, space, nobody cares how elegant the architecture is. They care about what actually changed. Less downtime. Massive improvement in output. All about outcomes. The fastest way to earn credibility is to tie your value to the one metric the operator already loses sleep over. If you can’t explain the before and after in plain operational terms, it probably isn’t real yet.
-
Jay Patil reposted thisThrilled that Loft Federal has won a seat on the SHIELD contract vehicle to support Missile Defense Agency with reliable and rapidly-delivered space capabilities!Jay Patil reposted thisLoft Federal is pleased to announce it was awarded a contract/s for the Missile Defense Agency Scalable Homeland Innovative Enterprise Layered Defense (SHIELD) indefinite-delivery/indefinite-quantity (IDIQ) contract with a ceiling of $151B. This contract encompasses a broad range of work areas that allows for the rapid delivery of innovative capabilities to the warfighter with increased speed and agility.
-
Jay Patil liked thisJay Patil liked thishttps://www.producer.ai/ Proud to have been an advisor on this project from the jump and to see it build rather stealthily with real users into an eventual Google DeepMind Acquisition. Lyria 3 has benefited immensely by the quality of the Producer.ai model - which will always be 'Riffusion' to early fans (turned team members) like myself. If you haven't checked it out - please do - one of the most forward thinking creator products out there. The 'Spaces' feature is absolutely best in class at the moment.
-
Jay Patil liked thisJay Patil liked thisAs we have been demoing our vertical AI platform, I love seeing when our clients' eyes light-up when they realize how powerful the tool is because of how well it understands their business. Thanks to Fast Company for including Crisp in their "24 technology trends to watch this year" piece https://lnkd.in/e6WsvGdX
-
Jay Patil liked thisThis is one of *the most advanced companies in tech* when it comes to sophisticated, "world class" use of AI. Incident have adopted cowork across the company for their employee's personal agents, and Credal as the cloud agent execution environment. This is the way!Jay Patil liked this“World class use of AI, across the company, on my desk by Monday at 9am” — Stephen Whitworth, CEO, incident.io, ~6 weeks ago*. No pressure then. *Actual words may vary; recollection poor. So what have we built so far? We started with commercial sales — reps selling a technical product to engineers, across dozens of deals at once, while our product team ships 50+ features a week underneath them. We bet on Claude Cowork, and built a single plugin that gives every rep two things: the best available knowledge as markdown, and the best recipes for success as skills. Our knowledge base is 1k+ markdown files — competitive battlecards, product positioning, case studies, pricing, the full product documentation from our core repo. It syncs automatically: a pipeline pulls from Notion databases, GitHub repos, and live website URLs on an hourly cron, renders to markdown, and commits. When a product marketer updates a battlecard in Notion, it flows through to every rep's plugin within the hour. We built over 20 skills. Deal research, call prep, call debrief, competitive analysis, business cases, CRM coaching, customer references, evaluation prep. Each is a markdown instruction file — a recipe that tells Claude how to approach a task, what tools to call, what evidence to look for, what format to use. We accept we can't control all the ways they'll be used: we hire great people, they will innovate and push the boundary. We want that. Our internal MCP server is a critical piece, and I’ve written about that already. We also wrote code to fill two specific gaps in Cowork. We built a telemetry demon, deployed on every laptop, watches for MCP and skill usage, ships to Cloud Storage, lands in BigQuery, and we track adoption in our BI tools. And we built CI/CD distribution: Cowork's organisation plugin system doesn't work for us, so a Kandji script on every managed Mac syncs plugin zips from a GitHub release every 15 minutes. Both are peripheral infrastructure — stuff that aims to be deleted once the platform catches up. Then there's automation. We’re adopting Credal as our cloud agent execution environment. We’ve built out a "delegator" system — scatter/gather with agents as the map function — that audits every active deal against our sales playbook. It’s cool. More soon. The team tell us they do in 15 minutes what previously took hours, which is the signal we need to know to keep pushing. We’ve noticed that the less scaffolding between the model and the work, the better. If you’ve been following along, what would you like to hear more about?
-
Jay Patil liked thisJay Patil liked thisRyboDyn has announced the initial close of a $10M Seed financing to accelerate our transition from foundational discovery into scaled platform execution and early therapeutic development. While advances in AI and protein engineering have transformed drug design, the industry continues to face a core bottleneck: target discovery. Many of today’s most well-known oncology targets are already crowded, limiting their impact across broader patient populations. RyboDyn is taking a different approach. By decoding the dark proteome, we are uncovering previously inaccessible biology and translating it into novel, disease-specific cancer targets. Powered by our RyboCypher™ platform and CypherAtlas™, one of the largest integrated dark transcriptome and proteome resources in oncology, we are building a new foundation for therapeutic discovery. This financing will support continued expansion of CypherAtlas™, advance our pipeline toward IND-enabling studies, and deepen collaborations with leading partners as we translate cryptic targets into new therapeutic opportunities for patients with limited treatment options. We’re grateful to our investors for their support as we continue to build. Read more in our press release: https://lnkd.in/gBZKMbSkRYBODYN ANNOUNCES SEED FINANCING TO ADVANCE AI-DRIVEN DISCOVERY OF HIDDEN CANCER TARGETSRYBODYN ANNOUNCES SEED FINANCING TO ADVANCE AI-DRIVEN DISCOVERY OF HIDDEN CANCER TARGETS
-
Jay Patil liked thisExcited to share this one. RyboDyn Inc. just announced their $10M Seed financing, and Swell is following on after investing in the pre-seed because the science and our conviction have only gotten stronger since we first backed Imad Ajjawi, PhD, MBA and Corey Dambacher . Cancer drug development has a target problem, not a drug design problem. AI is making it faster to design molecules, but the targets those molecules chase are more crowded than ever. The most well-known cancer targets each have 200+ drugs approved or in development against them. When the entire industry is optimizing against the same handful of proteins, you’re competing for smaller and smaller slices of the same patient populations. RyboDyn built an AI-powered discovery platform that decodes the 98% of human DNA that was written off as junk for decades, surfacing cancer-specific proteins that are druggable and invisible to conventional sequencing. Their proprietary atlas now spans roughly 1,000 tumor samples across 10 cancer types, with over 15,000 cancer-specific peptides identified. In breast cancer patients who lack the most common targetable marker, they found a novel target in about 45% of tumors analyzed, many of whom have almost no targeted options today. The team has already demonstrated the first-ever antibody drug killing of tumor cells through one of these hidden targets. That’s AI applied to biology in a way that actually matters in the world. Congrats to the entire RyboDyn team. Swell VC is grateful to be part of this from the beginning.Jay Patil liked thisRyboDyn has announced the initial close of a $10M Seed financing to accelerate our transition from foundational discovery into scaled platform execution and early therapeutic development. While advances in AI and protein engineering have transformed drug design, the industry continues to face a core bottleneck: target discovery. Many of today’s most well-known oncology targets are already crowded, limiting their impact across broader patient populations. RyboDyn is taking a different approach. By decoding the dark proteome, we are uncovering previously inaccessible biology and translating it into novel, disease-specific cancer targets. Powered by our RyboCypher™ platform and CypherAtlas™, one of the largest integrated dark transcriptome and proteome resources in oncology, we are building a new foundation for therapeutic discovery. This financing will support continued expansion of CypherAtlas™, advance our pipeline toward IND-enabling studies, and deepen collaborations with leading partners as we translate cryptic targets into new therapeutic opportunities for patients with limited treatment options. We’re grateful to our investors for their support as we continue to build. Read more in our press release: https://lnkd.in/gBZKMbSkRYBODYN ANNOUNCES SEED FINANCING TO ADVANCE AI-DRIVEN DISCOVERY OF HIDDEN CANCER TARGETSRYBODYN ANNOUNCES SEED FINANCING TO ADVANCE AI-DRIVEN DISCOVERY OF HIDDEN CANCER TARGETS
-
Jay Patil liked thisHonored to be moderating this upcoming panel on the state of the mid-cap GP Stakes market!Jay Patil liked thisMid-cap GP stakes are entering a pivotal phase as pricing, structures and governance models evolve. At the 2026 GP Stakes News Conference, Stefan Paulovic, co-head of our GP Solutions practice, will moderate a panel that will focus on where deal flow is heading and how buyers are underwriting growth in an expanding mid-market opportunity. Panelists include Ajay Chitkara (Bonaccord Capital Partners), Michael Lunt (RidgeLake Partners), Anthony Maniscalco (Investcorp), Michael Shedosky (Azimut Alternative Capital Partners). Request to join: https://luma.com/dpuyc419 #PrivateFunds #GPStakes #GPStakesNews
-
Jay Patil liked thisJay Patil liked thisI’ll be in DC this Wednesday and Thursday at the Satellite conference. Looking forward to connecting with space industry friends - message me if you’d like to chat!
Experience & Education
-
Swell Partners
View Jay’s full experience
See their title, tenure and more.
or
Other similar profiles
Explore more posts
Explore top content on LinkedIn
Find curated posts and insights for relevant topics all in one place.
View top content