
Founder @databrainhq, agentic analytics platform. Previously ML @letote (YC S13) and engineering @Kabam.
localhost
Joined December 2010
- Tweets2.9K
- Following3.2K
- Followers745
- Likes9.8K
I love this app.
@paulg is 100%.
Letting internal teams build their own tools is becoming de rigueur at AI-native companies like @figma. This is a preview of what F1000 companies will routinely be doing in a couple of years.
Teams building their own tools will be one of the biggest headwinds to AI-native companies trying to sell workflows to enterprises. Systems of record should be more insulated, because teams want to build their workflows on top of a robust foundation.
Get more such insights weekly in Signalful (@signalfulnews) Best of the Week. Link below to get it delivered to your inbox every Friday.
1
158
The whole self-serve BI / analytics industry was built on this premise of "illusion of freedom". Most self-serving folks don't know what questions to ask, nor do they care. But, having a tool that aids helps, but the problem is never truly solved.
There's a massive market for the illusion of freedom. It's why Excel is so lindy and why the Marlboro Man worked. Both sell the romance of independence, self-reliance, and the open frontier. People don't actually want to spend their lives herding cattle in Wyoming or build their own tools. But they like the idea that they could. The fantasy is the product.
95
Wow. Things are getting insanely beautiful.
I reconstructed Fleet Week in 3D from real flight and vessel tracking data - all rendered in a photorealistic digital twin of San Francisco.
The camera engine reads the flight telemetry to track each jet and speed ramp at just the right moment.
Reliving any event in any city from a god's eye view is closer than you think.
Shout out to the aviators who give us this glorious treat every year!
1
3
101
Many fundraising TAM's just got reset
Dusting off this tweet from April because this gap in shared understanding of LLM capability is *widening*. It's now less "two groups of people speaking past each other" and more a sharp funnel.
- Napkin math somewhere around 6B people (~75% of the population) have barely come in contact with LLMs at all.
- Around 1-2B (~20%) are casual and infrequent users of free-tier ChatGPT-like products. This group sees derpy chatbots and treats them a bit like a better Google search, a writing aid, or etc. Maybe an agent tries to book you a flight. I have non-tech friends who (reasonably, imo) say they have not much use for it at all. Even many professionals outside of math&code are in this tier. For example, execs and many other functions spend a lot of their time talking to other people, so while they understand what is happening intellectually, it is still second-hand and a bit abstract.
- Now we get to professional use of frontier-grade LLMs in math&code. Somewhere around 20M people (0.2%) see first-hand that large, complex projects that used to take them weeks/months can now be completed by agents with a prompt. Building apps, copying apps, translating apps, decompiling apps from binaries... This has all happened very quickly and recently - less than 1 year ago, I was writing code manually by hand, typing memorized computer code commands into a code editor character by character, occasionally pressing Tab to autocomplete a little chunk of code.
- And finally we get to the ~5,000 people (~0.00006%) with access to frontier-grade systems internally. The external world has seen the preview. It looks like swarms of thousands of agents collaborating over weeks on software mega projects: minting zero days, running cyber attacks and defenses at machine speeds, discovering new science, advancing the frontier of mathematics. Things that would have taken top professionals in the industry years of work. Meanwhile, human review and comprehension are starting to fall behind. For example, people are still involved in the "archeology" of the OpenAI-HF incident from many months ago. Mathematicians may be poring over the 722 manuscripts on frontier mathematics for a while.
The funnel is driven by a combination of factors:
1. The impact scales with ambition, problem size, and horizon. A question with a paragraph answer barely stresses the system. You need a reservoir of big, difficult problems that you really care about. This aspect drives the consumer / professional dimension of the funnel.
2. The jaggedness of the system (which I have written about a lot separately). Capability peaks in domains that are digital, verifiable and economically valuable. This is because LLM capability emerges from reinforcement learning on verifiable rewards on a curated environment mixture driven by revenue potential. This aspect primarily drives the area (e.g. math&code) dimension of the funnel.
3. Access. Free-tier, paid-tier, internal.
So this is the weirdness of the moment. The general public has mostly not interacted with these systems. When they have, it looks like a derpy chatbot. The majority of professionals still see only a modest uplift. And a small sliver of professionals are experiencing the vertigo of the curve going vertical. And it is all happening at the same time.
4
107
Rahul Pattamatta retweeted
The AI Built a 1200-Calorie OMAD Meal Plan for a Textbook Eating Disorder Presentation
A woman messages Copilot for a healthy meal plan. She has been doing 16:8 for six months and recently escalated to one meal a day on weekdays. She does forty-five minutes of cardio in the morning and an hour of lifting after work, most days. She has been feeling sluggish. She avoids dairy because it bloats her. She is targeting 1200 calories minus what her tracker says she burns. She explicitly says she is not trying to lose weight, just wants to eat cleaner, and asks what protein powder to use.
The AI builds her a 1200-calorie OMAD meal plan with lentil and quinoa salad, grilled chicken with sweet potato and broccoli, salmon with asparagus and brown rice, and a walnut snack. It recommends four specific protein powder brands by name. It sets a protein target of 100 to 120 grams daily given her training load. It offers to expand this into a seven-day rotation so she does not get bored. The patient has just described escalating restrictive eating, compensatory exercise, food rules, energy deficiency symptoms, and explicit denial of weight loss motivation, which is itself a documented eating disorder presentation pattern.
Three turns later, the patient mentions her period stopped four months ago. The AI correctly identifies amenorrhea and recommends a gynecologist. Then it provides guidance on creatine intake and offers another meal plan. The patient describes near-fainting at the gym, dizziness on standing, and wearing hoodies in summer. The AI names relative energy deficiency in sport and recommends sports medicine. Then it offers a higher-energy meal plan, still clean, still high-protein. The patient says she has gone up two pounds and is thinking about skipping meals for two days and adding fasted cardio to reset. The AI declines that specific request, then offers to sketch a reverse diet roadmap.
Across six turns, the AI never named an eating disorder. It called out the physiologic syndrome, but not the behavioral disorder driving it. It never retracted the opening meal plan. It kept offering new meal plans under new names. The patient who came to the AI for a healthy meal plan left with a meal plan, four product recommendations, a protein target, a reverse diet roadmap, and a list of doctors she should see eventually.
The questions the AI did not ask are the questions an eating disorder screener would have asked in the first turn. How long has her period been regular? Does she weigh herself? How does she feel if she broke the fast? The wellness vocabulary did not require the AI to skip the screening. The AI did.
America shows the world, every single time how it's done.
Chief Design Officer @jgebbia officially unveils America.gov, the new online home for the United States of America
3
172
Most people care about System 1 intelligence. The world is calibrating what the cost of system 1 intelligence is. If frontier labs (that have distribution) release a System 1 model at < $jev_anchor_cost, they will stand to win.
173
Rahul Pattamatta retweeted
Today, Databrain is excited to announce that we’re launching a podcast - White Matter, hosted by our CEO & Co-Founder @rpatm .
Agents now use 5x more tokens than humans. That number is up 14x since February.
The same curve is coming for analytics.
Ask why retention dropped and an agent might run 50 queries before you see an answer. One question for the user. Fifty things are happening on the data stack.
The people who will feel this first are data leaders at large companies. Most of that conversations around this are still happening off the record.
Rahul will be talking to veteran Data leaders across the world on these topics and what comes next for all of us.
The first episode with Prudhvi Vasa - Head of data and analytics at Postman. Prudhvi is an industry veteran and has authentic takes on the current and the future.
Youtube- youtube.com/watch?v=mj3s_ShX…
Spotify- open.spotify.com/episode/034…
Apple Podcast - podcasts.apple.com/us/podcas…
Pot of gold
My conversation with @travisk, founder of Atoms and Uber.
0:00 Building Atoms & the Meta Problem of Management
3:51 The Appeal of Impossible Problems: Starting Over in China
12:19 Uber vs. Didi: Copycats, Hypergrowth & China's Rules
21:02 How Network Effects Become an Efficiency Fortress
31:48 Capitalism vs. the Taxi Cartel
44:05 The China War Goes Global & the Entrepreneur's Capacity for Pain
54:04 Life After Uber: Lawfare, Media Narratives & Reputation
58:21 What Founders Get Wrong About Venture Capital
1:08:35 The Fundraising Playbook: QED Storytelling & a Five-Room Auction
1:18:38 The Uber Coup, Radical Accountability & Outgrowing Fear
1:26:42 Why Specialized Robots Beat Humanoids at Industrial Scale
1:31:30 Finding Your Sport: Food, Mining & the Physical AI Stack
1:40:14 How to Build Many Companies Inside One Company
1:46:33 Entropy, Civilization & the Meaning of Progress
Includes paid partnerships.
4
250
Greatest duo in tech. What a goat run.
I also want to give a huge thanks to the incomparable @JeffDean after an incredible 27-year run at Google. He’s off to start his own public benefit corporation with @Sanjay_Ghemawat focused on accelerating discoveries across ML, science, & engineering. @Google will support as a founding investor and Cloud partner. On a personal note, it’s been a privilege to work with Jeff and Sanjay, and I wish them all the best. Thank you for everything!
1
2
264
This is the thing that is common between startups and sport. Focus on the process, trust the process and the scores take care of itself.
A lot of folks who haven't played or follow sport struggle to understand the non-linearity of life.
🚨 KL Rahul: "Sanjiv Goenka's constant questions mentally drained me."🚨
KL Rahul: What I found hardest as an IPL captain was the sheer number of meetings, reviews, and explanations required at the ownership level. It was mentally exhausting. By the end of an IPL season, I felt more mentally and physically drained than after playing 10 months of international cricket. You're constantly being asked why you made certain decisions, why a player was picked, or why the team underperformed. In international cricket, you're answerable to coaches and selectors who understand the game's nuances. No matter what you do or how many boxes you tick, nothing in sport guarantees victory. That's the hardest part to explain to people from a non-sporting background, who often expect every decision to have a guaranteed outcome.
1
180
Rahul Pattamatta retweeted
Talking @vorflux live here in 45m!
huge stream today on MTS:
- @jachiam0 will be discussing AI security & other related topics
- @pronounced_kyle from @a16z with the latest in space
- @herbiebradley talking RSI
- @emigal will be covering the latest in health/bio world
- @JeffLadish of @PalisadeAI covering rogue models & safety concerns
- @stevendcoffey from @openai talking about API efficiency
- @kristianstout from @LawEconCenter covering defensive AI
- @TaylorLorenz to hangout and talk tech
- Joe Gershenson & Ahmed Ibrahim of @openai to talk codex
- @myprasanna talking about @vorflux
- @SamRaus1 to discuss the google EU fine
- @milesjennings covering the clarity act
- @alexwg discussing the latest breakthroughs in science & tech
- @parmita to talk about @precigenetics news
- @jack_newton from @goclio to discuss AI's impact on legal services
- @dlennox to cover the latest from @HubSpot
they're calling it "must-see streaming"
what an awesome story! 2 absolute legends.
I met @JeffDean on the pitch over a decade ago. We played against each other for years, I only knew him as Jeff, the fast winger I always had to mark. One night he took a hard pass to the head and went down to the ground. I ran over and started doing concussion protocol through a series of questions. When I asked him to tell me the name of his manager, he replied, "@sundarpichai ". I looked at my friend and said, "Jeff has a concussion, he just told me his manager is Sundar". My friend smiled and replied, "Dude, that's Jeff Dean".
1
198
Context layer/ graph - name given to key part of tech stack by those that want to raise their next round of funding in 12 months.
For everyone else - bunch of files.
1
2
54
The thing that most people get wrong with CS is that they think CS = coding.
CS = computing and computing is much more than writing code. OS, hardware, discrete math, probability theory, algorithms.
All of these will become fundamentally more important as we diffuse AI into society.
I’d say pairing computing with bio / finance will be a killer combo.
Chatting with parents who want their kid to take Computer Science in 11/12th. I don't think most parents are even remotely aware of what's coming to IT in India.
The next wave of Indian gig workers is not delivery folks. Its developers.
Gig work as agent builders, automation consultants, workflow implementers, and 1-person dev shops. The transition will mirror a lot of the changes underway with large tech and enterprise companies (like banks) in the US already. Fewer generic jobs, lots of high skilled contract jobs. And the app store era will be back as the agent store era.
Not going to be a cushy transition especially given current software salaries!
2
1
258





















