ceo of @runrehearsals / previously pm at google labs
New York, NY
Joined February 2011
- Tweets1.2K
- Following393
- Followers9.8K
- Likes8.8K
Pinned Tweet
Everyone’s hyped up about digital twins and simulation technology but nobody is building transparent benchmarks in the space.
Today, in partnership with @GoogleDeepMind and @kaggle, we’re launching the first public human behavior benchmark.
Why is this important? How did we make it? Read 🧵
Out of the numbers, there are some super interesting findings.
- If a streaming service hikes the prices by $4, AI thinks 43% of people would stay versus in reality 70% wouldn’t leave.
- AI thinks that 40% of people dislike logos on their clothes, but in reality, only fourteen percent of people have a problem with that.
- When it comes to renewing your car insurance, AI thinks 53% of people don't check out other prices, versus in reality, only 24% of people don’t.
There’s a bunch more of these findings on our website: runrehearsals.com/human-beha…
Find the methods, test set and try your model on @kaggle: kaggle.com/benchmarks/rehear…
I'm so excited to partner with @kaggle and @GoogleDeepMind on the first human behavior benchmark :)
Can frontier models actually predict how real people behave? 🕺
Today, in collaboration with @RunRehearsals, we are launching Human Behavior Bench, the first public benchmark that tests whether AI can understand human preference and everyday decision-making.
AI models have gotten increasingly good at coding, but knowing for whom to build is the next frontier of challenges for AI models to capture. Human Behavior Bench evaluates models on predicting real human behavior across health, entertainment, recreation, values, and consumption.
To build this baseline, we gathered ground truth data directly from real people and evaluated models across a wide range of everyday scenarios:
• Ground truth baseline: 1,244 US adults surveyed (June-Sept 2026)
• Some cohorts are traditional demographic indicators such as age, income, and location
• Other filters are behavioral e.g. GLP-1 patients and recent car buyers
• 750 cases total (149 public / 601 private) across finance, health, recreation, values, and consumption
• Scored using 1-TVD (total variation distance)
Base models still have a significant gap to close to accurately predict human behavior by cohort. LLMs’ predictions tend to converge towards the mean, while the true distribution of human preferences vary greatly.
Against a human noise ceiling of ~93, the current leaderboard scores are: Rehearsals’ Engram 1.0 (AI Twins) tops the leaderboard at 94.1%, followed by Gemini 4 Argon at 80.6% Claude Opus 5 at 80.2%, GPT-5.6 Sol at 79.5%.
ALT A leaderboard graphic from Kaggle titled "Rehearsals: Human Behavior Bench". The ranking chart lists nine AI models and their corresponding scores, measured on a scale from 70% to 85%. The models are ranked as follows: 1. Gemini 4 Argon with a score of 80.6% 2. Claude Opus 5 with a score of 80.2% 3. GPT-5.6 Sol with a score of 79.5% 4. Claude Opus 5.5 with a score of 79.3% 5. Gemini 3.8 Flash with a score of 79.2% 6. Gemini 3 Flash Preview with a score of 78.5% 7. GPT-5.6 Terra with a score of 77.8% 8. Gemini 3.5 Flash-Lite with a score of 75.9% 9. Grok 4.7 with a score of 75.2% The background features a subtle gradient with warm yellow and beige tones, and the source is cited at the bottom as the Human Behavior Bench Leaderboard on Kaggle.
What is 100% cash back? lol
We just launched a credit card that gives you up to 100% cash back rewards.
Credit card rewards are broken. A stagnating, ossifying industry has spent decades collecting fees while handing back perks that rarely deliver. Points that are hard to use. Lounges that are hard to reach. Fine print that works against you. Innovation stalled because it could.
So we built what they didn't.
Coverd flips the script. Up to 100% cash back rewards in a transparent matrix — a live, regularly updated grid where you can see exactly how to get to 100% cash back. No fine print, no guessing, full agency. Every swipe is an entry into the rewards pool, and you see it all happen in real time.
Grocery run. Gas station. Dinner date. You spend — we've got you Coverd.
This is real transparency and empowerment — no $900/year fees, no settling for B-tier lounges.
50,000+ people have joined the waitlist.
The doors are now open. The Coverd Card is live today.
Like + repost + comment "COVERD" — we're pulling people straight from this thread for priority access to our metal card program. #getcoverd #coverdcard #personalfinance
Readers added context they thought people might want to know
The Coverd card's base cash back rate is 0.5% per dollar spent, with up to 100% only available during limited 3-minute surge windows for specific categories and the first purchase in that window. coverd.us/rewards-matrix… coverd.us/rewards-terms.…
Thank you for the invite 🙏
So many great events in Boston for @Techweek_ last night. I’m elated at the turn out for the @speedrun insiders dinner; was an amazing chance to meet tons of techs brightest leaders and local entrepreneurs.
Come hangout with our team and Hume AI at the best NY Tech Week event
partiful.com/e/Bc7dJv7aMLLlE…
flying not business class to #google who else is going? – at San Francisco, CA
time to regulate emoji 👏👏
💊 Emojis used as coded language to promote illegal activities online?
Some platforms are now detecting emojis used as code for drug sales.
This is one of the key findings of the first EU-wide report on systemic online risks.
Dive in → link.europa.eu/CQhKGc
#DSAForReal
google is all in on agentic coding 🫡
Gemini app on Mac
This was also the first project where we fully used Antigravity end to end. Not as an experiment, but as a real part of how we built.
Proud of the team, proud of what we learned, and even more excited about what’s ahead
gemini.google/mac/
📷
7 companies are throwing 20,000x more money at AI than the world’s 7th largest economy 😂
Crazy time to be alive
We’re building simulated AI humans from talking to thousands of real people
Exciting conversation about @RunRehearsals, the future of AI and society with the one and only @Scobleizer
Now your AI agents can do customer research by using simulated humans that are based on real humans.
In other words, a new way of building your business is here.
@runrehearsals cofounder @liambolling gives me one of the most interesting conversations about the future of business that I've ever had.
Shows how things are changing due to AIs. A few years ago I never would have imagined that this would be possible (I did consumer research all over America).
it’s not “some customers” it’s all of the USA lol
