Vijay Krishnan

The Anthrogen team (Y Combinator S24) just launched Odyssey — a 102B-parameter protein language model and one of the most impressive examples I've seen of what tiny, focused teams can achieve in AI × biology. Anthrogen is a team of six, and they just raised seed funding after doing YC. You don't need $100M to build a foundation model today. They replaced self-attention with a new architecture and trained with a diffusion objective inspired by evolution. From Ankit Singhal's post "We introduced a replacement for self-attention, we call Consensus. Instead of letting every residue talk to every other at once (self-attention), Consensus encourages nearby neighborhoods to reach agreement first, then lets that agreement propagate outward over a sparse contact/sequence graph. You can think of it as repeated local “vote-and-average” steps that spread information reliably through the chain and its contacts—mirroring how structural changes actually ripple through a protein. We find consensus to be very robust to LR changes—huge when it comes to scaling models up and to encourage faster and more accurate coverage of the theoretical loss landscape. In the coming months, we have some more fun releases planned to show off our million-scale lab data collection efforts and a more in-depth examination of consensus!" Read about their launch here: https://lnkd.in/g4BdDqCg

270

6 Comments

The future of software is up for debate. In this very special first live recording of No Priors, Chirantan "CJ" Desai shares what stays durable in the AI stack and what has to keep proving value. Watch our full conversation, where he calls MongoDB “truly the only disruptive force” in a market that’s existed for decades.

226

19 Comments

While OpenAI signed $1.15 trillion in compute contracts through 2035, DeepSeek trained a frontier model for $6 million. This was 2025’s central question : are we building on bedrock or quicksand? The top 10 posts of 2025 examined some of these topics : Are we in a bubble echoing the telecom crash, or building the next internet? Do traditional exit paths still work when secondaries dominate & IPOs vanish? How do you design tools when the user is AI, not human? 2025 forced a reckoning with reality. 1. How AI Tools Differ from Human Tools (https://lnkd.in/d9qcnbyz) : I consolidated my 100+ AI tools into unified, parameter-rich interfaces based on Anthropic’s research. The counterintuitive finding : AI systems need complex tools with complete context, while humans need simple, chunked interfaces. Claude’s success rate approached 100% after the redesign. 2. Back to Text (https://lnkd.in/dWcy53fv) : How AI Might Reverse Web Design : I watched an open-source agent book flights by navigating airline websites, extracting data from visual chaos. If AI thrives on pure text, the future of the web might look exactly like it started : simple text, but for robots instead of humans. The better AI performs, the fewer websites we’ll visit. 3. Circular Financing (https://lnkd.in/d8-KMGZf) : Does Nvidia’s $110B Bet Echo the Telecom Bubble? : Nvidia’s vendor financing totals $110B in direct investments plus $15B+ in GPU-backed debt, 2.8x larger relative to revenue than Lucent’s exposure in 2000. But unlike the telecom bubble, Nvidia’s top customers generated $451B in operating cash flow in 2024. The merry-go-round has paying riders. Read the full post here : https://lnkd.in/d8Xw6tJt

43

24 Comments

The future of voice AI is unfolding on two fronts: 1) Speech-to-speech models that capture not just what we say, but how we feel 2) Developer-first “Voice in a Box” APIs that make conversational apps work As voice becomes a dominant interface layer, the infrastructure behind it needs to evolve fast. The demand is already here: sales agents, schedulers, care providers. Every founder building in this space is building toward a much bigger shift. “Voice AI in a box” platforms will make conversational apps and agents more personal and make developers' lives easier, like getting phone numbers, scaling calls, translation, custom-branded voices, voice cloning, and more. This area is poised to grow fast both now and well into the future as interfaces like avatars become mainstream.   If you’re building here, I’d love to meet! https://lnkd.in/dD9U3jpy

110

21 Comments

On the latest episode of Embracing Erosion I sat down with Suyog Deshpande, Co-Founder & CEO of Webless (a Fluvio Ventures portfolio company). Before starting Webless, Suyog spent years at Amplitude, Salesforce, and Samsara - shaping products and GTM strategy at scale as a product marketing leader. That perspective is now fueling one of the boldest bets in tech: 🏗️ rebuilding the web for LLMs. Instead of optimizing for clicks and SEO, Webless imagines an agentic web -where sites are designed to interact directly with AI models and agents. We dug into: - What an LLM-native web could actually look like - How companies can prepare to be LLM-ready - What metrics will matter beyond pageviews and clicks - Why safety and trust are core to agent-driven experiences - Lessons Suyog brings from “big tech” into this transformation If you’re curious about how discovery, trust, and value exchange will evolve online, this one’s worth your time. Link to the full episode in the comments 👇

9

1 Comment

India is not behind on talent. It is behind on risk appetite. Building LLMs or foundational AI infra is not about capability. It is about capital. Not just any capital, but capital that is comfortable with long timelines, unclear outcomes, and massive burn. Most of India’s funding still chases near-term traction. SaaS, fintech, consumer. Sectors where returns are easier to model. But AI infra does not play by those rules. It demands belief before proof. The Middle East launching a sovereign fund to accelerate AI through public private bets is not just money. That is intent. India will not build foundational AI by hoping startups will figure it out alone. It needs conviction at the system level. And a willingness to back uncertainty at scale.

56

10 Comments

Here’s a truth that often gets missed: becoming an AI-first company at the enterprise level is still really hard. It requires making big investments in R&D and specialized talent and betting on still-evolving frameworks and tools — and having the commitment to push the field forward yourself where the right tools or integrations don’t yet exist for your specific use case. As deploying AI and agents becomes table stakes in more industries, some companies end up hyping prototypes or easy use cases on earnings rather than building complex, self-improving AI systems to move the needle on productivity and profit. The task of separating these AI talkers from AI doers in public markets  – and leveraging AI to do it – is why my longtime partner John Hughes and I created the Swift AI Index as a fun side project last year. We just updated the index with five new stocks as part of a quarterly refresh; check it out: https://lnkd.in/eNBGjQzK

79

7 Comments

Vinod Khosla is considered as one of the most successful and influential venture capitalists ever. He's an indian-american billionaire founder of Sun Microsystems and Khosla Ventures, with investments like Square, Stripe, DoorDash, Instacart and more. Mario Gabriele's Lessons from Vinod: 1. You may be closer to a yes than you think. 2. Get the goose, not the golden egg. 3. Expect resistance to innovative ideas 4. Do not forget: your job is to sell. Top resources from Khosla: 1. Pitch the way VCs think: https://lnkd.in/dVQ95uc5 2. How to solve the Pitch Problem: https://lnkd.in/dFKdTABy 3. Plausible tomorrows 2035-2049: https://lnkd.in/dGQdtA4b

92

9 Comments