vas natarajan

2.2K posts

vas natarajan

vas natarajan

@vas

partner @ Accel. head cheerleader @ gamma, lovable, radar, ironclad, transcend, permitflow. formerly: frame (ADBE), spoke (OKTA), segment (TWLO), deepmap (NVDA)

iPhone: 39.877762,-75.240646 Katılım Nisan 2009
547 Takip Edilen3.8K Takipçiler
vas natarajan
vas natarajan@vas·
The opposite of product insecure -- you gladly build paths out of your product, but use that as motivation to be so good that no one takes the option. @Lovable
Tyler Bruno@tylerbruno05

people keep saying you can't leave Lovable: "you can't get your code out" → two-way git sync, full history, runs anywhere "your data is trapped" → real postgres export, migrations in your repo. or bring your own supabase from day one "you have to host with them" → frontend on netlify, cloudflare, wherever. or move the backend and keep the frontend. mix and match why would you though

English
1
0
6
580
vas natarajan
vas natarajan@vas·
Models frozen on historic global data can misjudge local truth. Agents climb the steepest hill in front of them, but it may not be the one the user wants them on. Now the interesting product frontier isn't feeding models fresher facts — it's drawing the right objective out of the user and packaging it up for AI. That's a super complex pathway to solve, but awesome work being done there @noah_weiss @nuance_ai @thinkymachines
Thinking Machines@thinkymachines

We're building AI that people and organizations can shape and make their own. AI should extend our will and judgment instead of neglecting it; enabling that is the technical challenge we are working to solve. thinkingmachines.ai/blog/the-futur…

English
0
2
6
2.6K
vas natarajan
vas natarajan@vas·
Agents can replace entire application stacks, but in many industries they're forced to be guests in existing systems. The @useagave team spent years reasoning about legacy application ERP and financial software -- the data models, the integration paths. Now their agents power the operations of some of the most critical commercial and infrastructure projects in the country. Growing like a (ahem) weed, profitable. And an incredible, hard-to-replicate moat.
Accel@Accel

Many of today's most interesting AI companies are working in the toughest possible conditions, with complex data, systems built decades ago, and almost no margin for error. @useagave is one of them. Tom Reno, John Zucchi, and Pooria Azimi are building for a construction industry that is still running on incumbent software, where every job is its own business and few processes have caught up with the modern world. After leading Agave’s seed round, we're deepening the partnership by leading their Series A. Read more from Accel's @Vas Natarajan ⬇️ accel.com/news/leading-a…

English
0
2
11
2.6K
vas natarajan
vas natarajan@vas·
Building a warm, trusted relationship with founders while simultaneously isolating/ tackling key diligence questions on a company -- that's the spell our late-stage team has mastered. Our IP is less in "a playbook", but in developing partners who can wield that delicate balance in today's compressed timelines. That's how you earn an outsized share of the best breakouts in each vintage. Proud partner of this team @Accel 🦾
Molly O’Shea@MollySOShea

NEW: Inside Accel's Growth Investing Strategy An exclusive look at 40 years of legendary deals From a 10% stake in Facebook to: › Nebius (PIPE, up ~13X) › Cursor (now a $60B SpaceX acq) › Cyera (highest-valued private security co) Portfolio includes Anthropic, Spotify, Lovable, Supabase, Vercel, Scale, Decagon, UI Path, Crowdstrike, Linear. With Growth Partners Arun Mathew, Miles Clements, & Matt Weigand, we cover: › Agent economy & real enterprise adoption › Inference buildout and infra as the rate limiter › Token maxing & where spend goes next › 3 trillion dollar IPOs & the race to $10T companies › Why security exploded after Mythos

English
0
0
16
3.3K
vas natarajan
vas natarajan@vas·
One of the more interesting parts of applied AI is defining the goal to iterate against. Code compiles — its a machine-verified check an agent can goal-seek towards. But most verticals lack that check; the goal is an affirmative email, a signature, a nod, sometimes undefined altogether. Great applied AI products seem to find that goal and pull it into the loop.
English
0
0
0
304
vas natarajan
vas natarajan@vas·
One of the coolest things about @perplexity_ai is how it connects search queries to coding opportunity. At one moment you're looking up USMNT match times, the next you've been offered a polished React dashboard to track ticket prices across marketplaces. It sniffs out latent intent and cracks it wide open.
English
0
0
0
178
vas natarajan
vas natarajan@vas·
With their prior company Sqreen, we saw what damage the @TolmoHQ team can do when the right team is pointed at a vexing security challenge. Round two will be more magical. We need our best and brightest helping to secure this new AI-driven gush of code. I love this concept of the "Production Knowledge Graph". Kudos to @caseyaylward @accel on seeing this trend before most, and finding the team to meet the moment.
Casey Aylward@caseyaylward

Congrats on coming out of stealth @TolmoHQ! My favorite thing about this team is that they demo something every time I see them. My second favorite (and arguably the more special) thing is that @pbetouin @JbAviat @arnaud_breton and @poledesfetes have come back together after their time @SqreenIO to rethink one of the most important problems in security. More below.

English
2
1
9
2.8K
vas natarajan
vas natarajan@vas·
Obviously we’re huge believers in the leading model companies. Where value scatters from there implies where other great investments will emerge. We have a few! @Accel @radixark @modal @nebiusai
English
0
0
0
224
vas natarajan
vas natarajan@vas·
This👇🏼 is *the* question to wrestle with. How will workloads rotate across proprietary frontier models, their lower-cost (but still proprietary) sibling models, open models, and what’s done locally on the edge. Enormous implications for where the value accrues. Connected here is a fast evolving pricing dynamic, the map is shifting given the early entrant subsidization going on. “Follow the tokens” works insofar as you believe prices are stable. But as intelligence gets way cheaper, volumes can grow enormously and value capture can still fall. Just “following the bits” in the fiber days would have led you to some fabulous busts. IMO local and self-hosted is a fun dark wrench to be thrown in. A properly scaffolded Ollama + GLM or Gemma4 + Mac Studio is sufficient for a lot of work. And of course Apple’s angling for consumer edge inference - with a lot of user context at its fingertips.
Aaron Levie@levie

One of the biggest questions in AI is how far behind open weights models remain from closed models at any given time. There are huge differences in market structures depending on whether open weights models remain 3 or 6 months behind, or if they fall behind by years. The answer to this will determine how the chip stack plays out, where inference can be run, what sovereign AI looks like, what happens at the applied AI layer, what the margin structure looks like in AI, how much companies can afford to spend on AI, and more. At the moment the open weights players appear to be holding up at keeping close to frontier levels of capability. Will be fun to see how this plays out.

English
1
0
1
655
vas natarajan retweetledi
Jalen Brunson
Jalen Brunson@jalenbrunson1·
Somebody take Mikal’s phone
English
1.3K
10.7K
183.7K
6.6M
vas natarajan
vas natarajan@vas·
The path from raw inputs to AI end-user value is long and constantly mutating. Every layer has its own supply-demand curve with mid-cycle attempts at reinvention (e.g. grid power > on-site turbines, copper > photonics, HBM > next-gen memory stacking). Yes, the aggregate AI infra curve is steep. But alpha lives in the component curves: the companies, their individual dynamics, how they're financed, where the market still underprices forward demand… There's no grand unified theory. But two axioms apply to each sub-market: 1. Cost of capital is a thing. You create equity value when returns on invested capital exceed financing costs. Growth funded below that hurdle destroys value, no matter how big the TAM. That's the gravitational force that pulls us back down from lofty "revenue multiples." 2. Cyclicality is a thing. The second the market senses overshoot, multiples collapse — and you can be left painfully holding the bag. But combing through this tangle is the fun part.
English
1
0
2
151
vas natarajan
vas natarajan@vas·
Two interesting and (converging?) paths in applied AI land. 1. Land with an AI-as-intelligence product that maps to use cases in deep knowledge work domains, where reasoning about primary sources can be challenging. Fast time-to-value, good ACVs. Rapid growth. 2. Decompose a business process and re-encode via AI. Grind through change management to automate workflow. Lots of multi-step processes and presuming there's low error tolerance, startups can differentiate on verifiable or traceable work. Not quite the growth rates of bucket 1, more "deliberate" growth -- but potentially long-term stickier. Former are moving in the direction of the latter (whether explicitly or not) -- JVs, acquiring customers or building pseudo-BPOs. Latter are building the operational IP to help customers "get to outcomes". Ultimate question is where equity value accrues. The pie chart near term may heavily favor type 1 companies. But may rotate over to type 2 in the medium-to-long term. Fun reality is the pie itself appears so large that both can win. But the slices themselves are changing texture real-time.
English
1
1
13
1.9K
vas natarajan
vas natarajan@vas·
@caseyaylward Even if you concede technical distinctiveness / superiority, still the ongoing ops question of getting the raw materials in place and sustaining those channels. Maybe not surprising - but just a different layer of risk we old SaaS peeps aren’t used to.
English
0
0
2
128
Casey Aylward
Casey Aylward@caseyaylward·
@vas Is the moat who can get more compute?!
English
1
0
2
582
vas natarajan
vas natarajan@vas·
Decompose most AI infra startups and the moat resolves to supply-chain mastery — not just access to components, but combining or operating them. From a cold start that's a difficult layer to accurately underwrite. Pull the thread and diligence runs into fairly opaque corners of the market.
English
1
0
7
1.3K

Keşfet

@Lovable @noah_weiss @nuance_ai @thinkymachines @useagave @Accel @perplexity_ai @TolmoHQ