@anurag_singhal
Account based inUnited States
About this account
- Account based in
- United States
- Connected via
- United States App Store
Account-level information from X, not a live location or the device used for a specific post.
Founder of Quattr Inc, the AI Search Visibility Platform. We unify your data, content & linking strategy to get you cited in AI answers, not just ranked
Palo Alto, CA
Joined May 2009
- Tweets392
- Following610
- Followers179
- Likes77
What is most refreshing about @CompleteSkeptic and team’s Jev going viral this week is that we are not getting flooded by posts about videos, images or answers, AI slop inside. We are seeing real builders sharing what they are now able to build with AI - with faster cheaper better decisions at scale. Some of these demos are cool but impractical (today) but many are truly promising and ground breaking.
I was so wrong about where LLMs can keep taking us. Congrats team @typesafeai - game (Jev)on!
My X feed got hijacked all day today by @CompleteSkeptic and @typesafeai’s awesome Jev launch. Can someone please catch me up on what else happened today on X?
This launch looks promising. It has the vibes of a truly refreshing take on AI since LLMs took the world by storm. The economics are wild.
After co-inventing ChatGPT, I kept asking myself: why have superhuman chat models not led to AGI?
I’ve spent the last 2 years in stealth building a new way to train models (RLCD), and a new type of frontier AI model that we are releasing today: Jev
• 20-200x faster
• 40-400x cheaper (w/ output tokens free)
• Frontier composable intelligence optimized for decisions
AFAICT the shortest path to AI-based economic revolution
Thanks for a great intro @kunchenguid to your awesome agentic engineering workflow and set of tools that make it all possible and easy to setup.
My best takeaways or the Kun tricks I really liked:
👉 Use /calm to prevent getting into the habit of gazing an agent ramble tool calls as it works. Off by default. Small changes and good habits go a long way.
👉 Multi-harness setup: Delegate model + harness choices to a policy, maximize your usage across accounts
👉 Project ledger maintains preferences for key decisions such as no-mistakes or yolo workflows before PR is merged based on risk.
👉 Lavish acts as a richer decision board vs AskUserQuestion or harness specific ways to respond
👉 Firstmate can create secondmates to delegate multiple sessions across the same project. Free organization a la Claude projects
👉 Single tmux UI to manage work across macbook and mac mini.
There were a lot more goodies in there. But overall this session is gold and reminded me of one of @clairevo's podcast with @ryancarson a long while back when cursor rules used to be the thing.
finally found time to make a new video! youtu.be/MSbacZ99E14
this time i'm sharing a slightly more advanced agentic engineering session, focused on high throughput multi-tasking
hope you find helpful!
QM might do for raising agents what @ycombinator SAFE did for raising funds
(And most traditional investors still hate SAFE, frontier labs might hate QM)
And if MCP is the new HTTP, QM might be the new chromium.
We’ve decided to open-source a multi-agent harness we use internally at YC.
We call it “QM” and it’s meant to be easy to customize, like Hermes or OpenClaw, but useful for a whole company. We use it across accounting, legal, events, and engineering (including building QM itself!).
The whole project is under an MIT license. It is cloud-first and has Slack and web UI natively.
Anurag Singhal retweeted
An internal version of Astra, @OpenAI’s next major model family, solved 10 major open problems in mathematics, quantum complexity, and theoretical computer science.
We believe it will be a major step for scientific reasoning. openai.com/index/ten-advance…
X stopped penalizing links in the post. But @nikitabier that’s not enough.
X needs to truly solve the poor user experience by automatically surfacing the links OPs bury in the thread out of outdated fears. That will greatly improve X user’s experience. Because it is still rare to find them in the main post as the damage is deep.
New essay: The Apprenticeship We Automated
Agentic AI doesn’t just automate the work. It can automate the error exposure that turns work into judgment.
The question I now ask before I hit enter:
Is this difficulty tax, or tuition?
linkedin.com/pulse/apprentic…
Anurag Singhal retweeted
JUST IN: OPENAI HAS CONFIDENTIALLY FILED FOR AN IPO
"We recently submitted a confidential S-1. We expect it to leak so we're just announcing it. We have not decided on timing yet; it may be a while because there are things we want to do that are likely easier as a private company."
The numbers:
Current valuation: $852B
Weekly active ChatGPT users: 900 million+
Total funding raised: $180B+
The timeline: OpenAI is targeting Q4 2026 at the earliest. Goldman Sachs and Morgan Stanley are the lead banks, the same two at the top of SpaceX's filing.
Anurag Singhal retweeted
Yep, it's in Canary now. I just tested it and it drove me directly to AI Mode -> Chrome tests redirecting searches to AI Mode instead of Google Search results
"An experimental flag in Chrome Canary reroutes address bar searches to AI Mode. When enabled, search queries typed into the omnibox open an AI Mode conversation instead of the standard Google Search results page."
windowsreport.com/chrome-tes…
Sycophanoia (n.) — the creeping suspicion that your AI is still flattering you even after you told it not to, so that even its honesty reads as suspect.
I'd treat this as a major event in #AI that tilts the balance towards #Anthropic winning the AGI race. Anthropic recognizes the importance of architectural primitives like MCP the best and I hope to see @karpathy accelerate that lead with his first principle research and thinking.
Quattr customers rated us #1 for Results, Usability & Relationship in Answer Engine Optimization (AEO) on G2 Spring 2026 — earning 43 badges across four categories.
businesswire.com/news/home/2…
1/ Last week I shipped my first production code in 20 years.
Then a second feature. Same week.
I'm a CEO. Haven't touched production code since my engineering days. Both would've taken months — meetings, Slack threads, Jira tickets, sprints.
It happened over a weekend with #ClaudeCode
9/ The fear shifts.
It's no longer "I can't build this."
It's "if I don't build this now, someone else will — tomorrow."