Jay Patil

WHY VC? Why will @swellvc end up being my life's work? It's because all my experiences have shaped me to become the person I am, and armed me with the skills/strengths which I believe are crucial to success as an early-stage investor

@datarade @TaskletAI

Like and Reply with your favorite AI tool that you think is a bit obscure.

Alex, Pierre and the Loft team have always been years ahead, quietly building next gen capabilities before the hype cycle catches up

@MichaelLatman I have multiple agents, and I use on desktop and mobile web. Would be nice to not have to remember what the last used agent was. And not having a way to route iMessage to a particular agent is a problem

@jaypatil Routing is tied to last active agent right now. Curious about your use case - are you switching between agents mid-conversation, or wanting to invoke a specific one from scratch?

the exact problem @credal_ai is solving in production today: permission-sync'd agents across Slack/GSuite/Salesforce/etc, auto PII redaction, human approvals on risky actions, full immutable logs Multiple specialized agents, governed and compliant.

@jonnydimond @TaskletAI Is there shared context across chats/agents? And can we invoke tasklet agents within shortwave AI chat?

At @TaskletAI we’re building agents that do real work - not just chat. A big challenge: how do you architect a long-running agent that doesn’t forget and can reason over massive amounts of data? Here are a few tricks we use 👇 1/ Dynamic, infinite context We want users to continuously tweak an agent in a single “infinite” conversation. But not all context is equal: User intent > tool output. Patterns > individual actions. So we built a context manager that compresses history dynamically, preserving what matters (e.g. “this happened 100 times”) while keeping context cheap and manageable. 2/ JSON-only tool results Every tool returns structured JSON. Yes, it costs more tokens - but we get 3 big wins: • Safer by default (JSON resists prompt injection surprisingly well) • Can be be processed with code - not just vibes • Aggressive truncation, with on-demand field lookup when needed The agent sees structure, not blobs of text. 3/ Everything is a file This is the glue that makes the 2 other tricks work. The agent has a virtual filesystem. Every tool result - even binary data like images - becomes a file the agent can read with code. That means the agent can read, transform, query, and compute over real data - without using many tokens or hallucinating. Example: the agent can average by actually iterating over result JSON and doing math — not guessing. We’ve learned a lot building this — I’ll share more as we go.

@datarade Yes

I have a private thesis on power grid infra I'm sharing with people, if you're interested in reading it, reply to this. Would love to get the hive mind looking at it.

@_onlyscott Sterling

89% of football fans will fail this 🫣 Guess the player?

@hthieblot AI could help filter through the noise. But for now, signal lives and dies with human interactions

What if investors had to fill out an application to invest in your startup?

@hthieblot Love the sentiment but filling out an application both ways is DOA. Adverse selection is real. The best of the best aren’t filling forms. Ever. ☠️

@hthieblot swell.vc

Who are the investors still investing in december? Raise your hand

@datarade Can listen to this while he’s driving podcasts.apple.com/us/podcast/how…

Have a friend who's a former pro footballer but had an injury. He makes ~5k pounds/month after taxes doing Bolt/Uber and lives ultra cheap with the parents in London, cooks at home, etc... What's the best book for him to read to gain financial literacy?

@jordihays I miss the simple days of old fashioned lies and sweatshirts

Rage Baiting is for Losers Yesterday, YC announced Chad IDE aka “the brainrot code editor.” Chad is an AI code editor that allows you to gamble, watch TikTok, and use dating apps while working on coding tasks. Their launch rightfully got a lot of attention. On one hand it’s funny. On the other hand, what are we doing here and why does this belong on the official YC account? To understand Chad IDE, Cluely, Icon, Friend, and the new class of Gen Z startups, you have to understand the online environment these founders grew up in. If you grew up on the internet and studied how and why certain people would regularly go viral, you know that making people mad has and always will be a highly effective way to get attention. The feedback loop is simple: 1) make something (product or ad) that makes people angry; 2) people comment/ share/ dunk; 3) because feeds are optimized to show posts with high engagement the most, you get more reach. Rage baiting for commercial purposes was pioneered by course bros. People like Tai Lopez realized that making the masses mad was an effective way to drive course sales. They could flaunt Lamborghinis, make a bunch of people angry, and as long as a handful of people found their way into their course, it was a viable, repeatable strategy. Historically on X, rage baiting was a marketing strategy, not a product strategy. Accounts like @sweatystartup frequently post things to get an angry reaction and subsequent reach, but behind the scenes he's always been running a normal commercial real estate fund. In 2025, rage baiting has become a product strategy. Cluely started as an app for cheating on coding interviews. Chad IDE’s only known differentiation from the other hundred AI native IDEs is that you can gamble and swipe on dating apps in it. The rage bait is sitting at the product level now. It’s becoming clear that while rage bait might occasionally work as a marketing strategy, it really should not be employed as a product strategy. Running a successful VC-backed company requires you to build a coalition of people that want to see you win. Getting media, investors, talent, and customers on your side is not an easy task. Rage baiting (whether at the marketing level or product level) is the most effective way to get people (who could be potential investors, customers, or team members) to actively pray for your downfall. YC has long provided some of the most durable, high quality, generalizable advice for startups and I believe it has had a tremendously positive impact on the companies that go through YC and even those that don’t. Launch now, make something people want, do things that don’t scale, ignore your competitors, etc. As someone who believes that YC is one of the most important and influential institutions in tech, I believe it might be time to include this in their list of essential startup advice: “Rage baiting is for losers.”

@levelsio @mirandanover building fort.cx

Fitness trackers having a bias for cardio and a bias against lifting weights may be partly why running clubs, marathons and Padel are now so popular It doesn't match with what the science says at all though: for a healthy long life you need both strength training and cardio A healthy body has strong muscles AND a strong heart This may also be one of the first times where algorithms are influencing health trends en masse, and possibly incorrectly

When Rusty and I started investing as a team 14 years ago, we decided to put two constraints in place to stay disciplined - fund size (stay small, laser focused) and check size relative to fund size (high conviction, concentration)

@typesfast Met a retired Amtrak conductor who swears by the theory that the Amish grow weed (a cash crop) and the women use Amtrak to deliver the goods across state lines

My friend says the Amish are super rich. They never spend any money. Grow all their own wood. Don't pay their children any wages to turn it into furniture, sell for inflated prices to tourists. Just stacking cash.

@DSox 👍🏽

Coming this fall.. If Family VC hosted a few coffee/bagel meetups in NYC for: 1. Family offices (active in VC) 2. Founders 3. VC managers Would you be interested in attending? * Reply or react to this post to lmk. *only if you are one of the three personas mentioned.

@IlyaAbyzov Hey Ilya would love an invite

Working on Torch 🔥 Unified health record + LLM in one iOS app. Syncs records from hospitals, labs, Function, One Med, PDFs etc. Makes it simple & fast to get all your health data in one place + get LLM help making sense of it. Reply/RT for TestFlight invite. More below ⬇️

@hthieblot Yes

Founders are over VCs being OOO all summer. Who’s still writing checks right now? Reply here.