
I don't blog often, but when I do, I buy the domain names. 😜 TheNewMoats.com UnitOfValue.com VCrisks.com TheOnDemandEnterprise.com EvolutionOfCloud.com CastlesInThe.Cloud OpenSourceVsCloudCastles.com TheNewNewMoats.com
✓Restless. Irreverent. Greylock GP. @GreylockVC

I don't blog often, but when I do, I buy the domain names. 😜 TheNewMoats.com UnitOfValue.com VCrisks.com TheOnDemandEnterprise.com EvolutionOfCloud.com CastlesInThe.Cloud OpenSourceVsCloudCastles.com TheNewNewMoats.com

Scoop with @rogoswami - Nvidia board is divided into factions, with one three-person faction vehemently against a Jensen Huang return. semafor.com/article/11/17/20…

The industry needs to move beyond long-lived API keys. There's been many recent examples of API keys being used as stepping stones in elaborate multi-step cyberattacks Excited to be launching support for this alternative alongside other great companies building with Claude

What if you could simulate how your customers will react to a price change, or how the largest IPO in history will trade? That’s one of the promises of AI-driven simulation, and the subject of our latest episode of @GreylockVC Change Agents: I sat down with @simile_ai co-founder and CEO @joon_s_pk to discuss the current state of simulation, and what the future could look like. We covered how Simile thinks about building confidence in simulation models, scaling to society-wide simulation, and how accurate simulations will help enterprises (and beyond) make better decisions. 00:00 Opening 01:02 Creating teams that fit the problem 03:11 Hiring people with contradictions 09:34 A research bet that paid off 12:05 The gap between what people say and do 14:59 Four stages of simulation 19:02 How customers use Simile 20:57 Getting the information to simulate an individual 26:44 From reactive to proactive simulation 32:50 Research vs product focus Some of what we discussed: -Confidence was important to Simile’s customers, who wanted to know how likely it was that a specific simulation was right. To solve this, Simile trained another model, jointly trained with their core model, to evaluate the core model’s confidence for any given scenario. It’s a main feature of Simile today and a big research bet for Simile that paid off. -As companies get more used to simulation, it will evolve from a reactive to proactive tool. Right now, most of Simile’s customers want to answer specific questions or test specific scenarios. In the future, however, they might just come to Simile with a desired outcome. It’ll be the simulation’s job to tell them how best to get there. -Societal simulation is the ultimate goal, but it doesn’t have to happen all at once. The most complex stage of simulation is societal simulation, in which multiple agents interact within a rich environment. It’s a big task, but getting there doesn’t mean immediately simulating the entire world and all 8 billion people in it. Instead, we can get there in a series of bite-sized milestones that still have big market impact. Thank you @joon_s_pk for joining!

Incredibly proud of the @Onehousehq team. The Gartner list is mostly hyperscalers and very large enterprise software companies, plus two mid-sized startups and two real startup challengers. If scored on technical capabilities alone, the leader quadrant will look quite different.

The name of the game for cyber tools is now customization. Each company has its own unique infra, data, and security practices. This means the old Swiss-cheese approach of layering generic security products and hoping they cover every gap won’t work anymore - AI-based attacks can easily discover the loopholes. Cyber products are already notoriously hard to integrate. Even best-in-class products are frequently missing integrations, need extensive tuning for false positives, and are built around assumptions of standardization, whether in security practices, workflows, or data formats. As the cost of building bespoke software collapses, security orgs will face an increasingly tough decision in the coming years: to buy, or to build? Products that survive will likely combine foundational infrastructure that's hard to replicate (endpoint/IAM/SIEM, etc.) AND deep customization where it matters: detection engines/rules, policies, integrations, and business logic around workflows.

🏆 Onehouse earned an honorable mention in Gartner’s 2026 Magic Quadrant for data lakehouse platforms. And we’re the only full-featured lakehouse on that elite list that can deploy anywhere. It’s great to see Gartner recognize the full breadth of today’s Onehouse platform. We started small. Now Onehouse is a fast, full-stack lakehouse: ingestion, Spark/SQL pipelines, cataloging, table management, compute, and low-latency serving. But we’ve stayed stubborn about one thing: customers should own a single open copy of their data, use the engines they want, and run the platform where they want. That flexibility, paired with a fully managed lakehouse, is exactly the vision we’ve been building towards for years. Gartner captured it in a few crisp sentences. Excited for where this goes. More to come.

And that’s a wrap on SFSRG 2026! Huge thanks to @AryaTschand and @aaravwattal for leading our largest session yet on one of the most important topics in technology today. Can’t wait for many more- dm me if you’d like to lead a future session!


Today we’re introducing ParseBench-2 - our most ambitious document parsing benchmark to date. It is the only benchmark that provides both extreme granularity into every single type of production parsing error, while also making sure that metric scores are highly aligned with human judgment. We do this by painstakingly assigning severity scores to 330+ error types, and tuning each metric to match the severity scores. Existing benchmarks either use brittle deterministic metrics (e.g. OmniDocBench) or LLM-as-a-judge (ParseArena), which can help recognize equivalent representations but are coarse/stochastic/subject to bias by the prompt+model. ParseBench-2 also contains: ✅ 4.3k+ document pages (double from ParseBench-1). ✅ An extra dimension (forms) ✅ Much better ground-truth across all dimensions compared to ParseBench-1. This is the next evolution of ParseBench, our hugely popular benchmark that has 50k+ monthly downloads, over 150+ OCR solutions benchmarked, and has been adopted by companies like NVIDIA, Cohere, Aleph Alpha, Perplexity, and others. Alongside ParseBench-2, we are also releasing OCRia-1.5 - a new family of document parsing models that powers all our modes in LlamaParse. They represent a massive step up in capabilities from our latest release, and are state of the art in parsing complex tables, forms, charts, formatting, and assigning bounding boxes to complex layouts. Come check out our blog post, technical report, and website. Blog: llamaindex.ai/blog/introduci… Technical Report: llamaindex.ai/files/parseben… Website: parsebench.ai/ If you’re interested in trying out our OCRia-1.5 capabilities, come sign up for LlamaParse: cloud.llamaindex.ai/

Where will durable value land in agentic coding: the model, the harness, or the gateway? Join Greylock Partner @jerrychen as he moderates a discussion with leaders from Microsoft AI, Nous Research, and OpenRouter at Microsoft for Startups Innovation Basecamp on November 18 in San Francisco. Register by November 12: lnkd.in/gW2hHsXB Free for founders. No Ignite pass required.

Congratulations to @SevenAICyber, @cogent_security, @CylakeAI, @Netic_AI, @parallax_ind, and @Tenzai_Labs on being named to The Information's 50 Most Promising Startups of 2026. 🎉 Proud to partner with these exceptional founders and teams from day one!

Ok but actually our vibe coded booth looks sick — come check us out at @aiDotEngineer CODE in SF this week and see a demo

This is what I get for vibe coding our booth design

just ask your agents what they prefer, slack or ando?

Announcing Dagger for GitLab. Run the same checks before and after push. See results in your terminal and merge request. Cut out the middleman: just GitLab and Dagger Cloud. Stop wasting compute: checks are cached, and you pay for exact usage, not machines. dagger.io/blog/dagger-for-gi…