The New Stack
Microsoft's new open source SDK pairs with Replit to push vibe-coded apps past the prototype stage and into enterprise production — with security and governance baked in by design. https://thenewstack.io/microsoft-build-2026-rayfin-replit-vibe-coding/?taid=6a1f6ea62b694b00012f84ee&utm_campaign=trueanthem&utm_medium=social&utm_source=twitter
elie
microsoft MAI tech report is a gold mine, one of the most transparent for a model at this scale.
this model uses zero synthetic data or distillation from previous models. this means reasoning, agentic behavior, tool use are all learned fully during post-training with no cold start. bold choice that makes it harder and requires more iterations to reach sota, but you get FULL control over your model series and it proves they are serious about being a frontier lab.
the tech report is insanely detailed and precise about numbers. to give an example, they give the exact MFU across all the iterations of the model, with the exact changes etc. they also share the full scaling ladder recipe, to my knowledge this is the first time i've seen this in a tech report at this scale
let's look at all of this in this likely very long thread 🧵
Mustafa Suleyman: Super excited to announce seven new world-class MAI models today. They represent what we consider a new era in AI designed to keep you in control and on the frontier.
First is our text foundation model, MAI-Thinking-1, exceptionally strong on reasoning and SWE tasks.
- It’s a
theUSshould lead on AI by continuing to develop the very best models, making sure they're safe, and getting cyber tools into the hands of trusted defenders.
the new EO gets the balance right.
Build and launch apps to your team, using Codex:
OpenAI: Building apps has never been easier.
With Sites, Codex can turn your work, ideas, and plans into an interactive website or app your team can explore, use, and share with a URL.
Rolling out to Business and Enterprise plans, before expanding more broadly.
💥Susan Dyer Reynolds🗞️
“Was the role of a traditional journalist to be compassionate? That’s a little hazy, but I think [it is] the role of a human, certainly.” https://thevoicesf.org/dion-lim-every-day-is-a-new-adventure/
Saining Xie
how does the brain build and track an internal state of the world from (possibly incomplete and noisy) visual observations?
i believe visual state tracking will be the grand challenge for vision in the coming years, and i hope this benchmark can be a useful starting line. enjoy!
Sihyun Yu: Can MLLMs actually track what's happening in a video?
Introducing VSTAT 🎯, our new benchmark for visual state tracking.
The tasks are simple: count cups, read typed words, count page flips. Humans solve them easily. MLLMs don't.
https://vision-x-nyu.github.io/vstat-site/
🧵 [1/11]
Suhail
1/ it all started w 2 8xB200s
excited to be back in the game again
Matt's guide is great but this line is pretty funny:
"They're so lit up by being able to build anything that they don't do anything else. Then they launch and there are no users." 😆
Matt Van Horn: http://x.com/i/article/2061440101411102721
T Wolf 🌁
Saikat Piker the carpetbagger is cooked. There is some justice in San Francisco.
Decision Desk HQ: DDHQ Race Update (est. 37% in): California US House 11 Primary
Scott Wiener (D): 37,562 (43.4%)
Connie Chan (D): 24,276 (28.1%)
Saikat Chakrabarti (D): 11,689 (13.5%)
Follow more results here:
https://votes.decisiondeskhq.com/races/251780
Rohin Dhar
I think what often gets missed
Is Y Combinator moving to SF in 2023
May be what ends up having the biggest economic impact on San Francisco eventually
Just a machine gun of billion dollar companies being produced quarterly in SF, some that will grow much much bigger
Faraz Khan: @pitdesi It probably was over if it wasn’t for OpenAI / anthropic.
otterboyinfrisco🏳️🌈
Love to see the Peskin-style anti-housing, anti-change anti-small business candidate get absolutely SMOKED by @scsherrill !
We're entering the golden age of just-in-time software
Garry Tan: http://x.com/i/article/2061176923531984896
Your whole business laid out on a canvas.
Manny Bernabe: Replit keeps blowing my mind. I'm working on something new, and in the same project I just keep coming up with new things to branch off and start building. It is crazy.
I'm excited to show you what we have for you later on this week. Stay tuned. 🍿
Spock Woz
garry tan is so right about not building massive rails factories for agents but nobody talks about what actually goes in its place
after building this way for a while the shift is actually super simple
1. your backend code should just be dumb hands and feet. no complex business logic, no nested if/else loops trying to predict what the model will do. just clean deterministic apis, db reads, auth, and sandboxes. the plumbing.
2. all the actual brains and workflow procedures live in markdown skills. the first time an agent solves a weird problem, it takes a minute. but instead of throwing that away you freeze the procedure by stripping out the specific data. next time someone asks for the same shape you serve it instantly and deterministically. zero agent latency, zero model cost.
3. and the golden rule for keeping the agent from burning your house down is that you never trust its self report. if the agent says tests passed or the write succeeded, you don't believe it. you rerun the check in your dumb code. you let go of control on the way out but you buy it back on the way in.
build the harness, not the factory
Garry Tan: We're entering the golden age of just-in-time software
💥Susan Dyer Reynolds🗞️
Two candidates — @saikatc claims to be a socialist and @TomSteyer might as well be — put millions of their own money into buying their elections and it looks like both will come in third. Think of all the social causes they could have helped if not for their greedy egos.
Mark Fabela: Internal polling was slightly off.
Thank you @ericries for saying what needs to be said and pointing the way to longer term companies focused on real missions instead of shareholder over everything
Dmitry Birulia — e/acc: Thank you @ycombinator and @garrytan for sending @ericries’s book INCORRUPTIBLE 🫶🏼
GBrain is the agentic swiss army knife for retrieval and memory
ani: so i wrote a session warden. small script that sits on top and watches your agent sessions.
catches crash loops before they spiral, kills zombie processes, and uses @garrytan's gbrain to shove context back in when a session restarts. agent comes back and just picks up where it
Frank Smith
I still hold that @MattMahanSJ was the best candidate for CA Governor. Alas, he entered the race late, and most Californians keep voting along party lines regardless of results.
As centrists, we need to keep pushing the movement behind Matt into the mainstream. These are winning ideas, and they point to a different way of doing politics:
1. Zero ideology, outcomes first
2. Get government out of the way where it’s a hindrance
3. Clear accountability for results
4. Try innovative approaches, fail fast, iterate
This is how government should operate in the 21st century—and how we restore faith in it.
Mayor Matt Mahan: Thank you, California!
probably the best reward function for reasoning efficiency i've seen
elie: length penalty is very elegant and simple tbh
Kane 謝凱堯
Congratulations to Saikat for spending $10M of his tech and private equity fortune pretending to be a socialist candidate only to be shellacked into a distant third place.
Saikat Chakrabarti for Congress: Our volunteers and field organizers are currently knocking 5,000 doors an hour to make sure everyone votes today.
This is the biggest field campaign in San Francisco history.
tfw you land many PRs and have many more to go before you sleep
Thank god Phil Kim beat the insane woman named Brandee who literally doxxed my home address and made me sell my home and move my family to a new one
💥Susan Dyer Reynolds🗞️: Wah wah Brandee Marckmann barely made a blip. She can take her spiderweb and go home.
codex for computer work is growing very fast
OpenAI Newsroom: Codex now has more than 5M weekly active users.
But the bigger story is what people are using it for: not just writing code, but getting more work done across research, analysis, content, and operations.
Our new report on how Codex is becoming a productivity tool for knowledge
roon
people have to build much faster software and interfaces so as to not Amdahls law the increasingly powerful and increasingly fast models
Neetu Arnold
“Admissions offices have fallen victim to ideological capture”
I often hear people say both grades & tests should matter. In the early 2010s, admissions officers seemed to agree
But over time, many believed grades alone were superior, casting aside the valuable info from SAT/ACTs
Wall Street Journal Opinion: The University of California needs the SAT back. Even the overwhelmingly liberal Berkeley faculty are fed up with the admission of unprepared students, write Svetlana Jitomirskaya and Zvezdelina Stankova
https://on.wsj.com/3RHHbFo
Time to short Uber and long DoorDash
Negligible Capital: *UBER SETS $1,500 MONTHLY CAP ON SOME AI CODING TOOLS FOR STAFF
$UBER officially reeling in the Claude budget after blowing their AI budget earlier this year.
Undoubtedly more companies to follow
Y Combinator
The next big social platform won't start with photos or videos. It will start with games.
http://Playabl.ai is the TikTok for user-generated games: anyone can play, create, publish, and monetize.
In 5 days, they hit 1M organic plays across 3,000 games.
GBrain SkillOpt now has 4 E2E evals that verify it working
https://github.com/garrytan/gbrain-evals/blob/main/docs/benchmarks/2026-06-03-skillopt.md
You know what to do 🗽
Peter Yang
My AI skills now grade themselves and get better the more I use them.
My new tutorial walks through exactly how to build skills with:
→ An eval loop to have AI fix its own mistakes
→ Memory so the skill improves over time
Skills are honestly incredible for encoding your knowledge and taste, and I can't get enough of them.
📌 Watch now: https://youtu.be/uT3EQPVIEb0
I have been thinking it's going to be the AI Harness Wars of 2027...
But after that comes the Frontier Labs vs All Software Companies war of 2028
Ankit Gupta: so, now that Codex Sites is out, will lovable have a higher or lower ARR in 1 year?
▲ + ❄️
Generating frontends on top of your business data is one of the killer apps of coding AI.
The genie is out of the bottle. Never going back to clunky and rigid dashboards.
@vercel already ran on Snowflake. But with @v0 and @nextjs, we’re now getting 1000x the value.
v0: v0’s Snowflake integration is now in public preview.
Prompt v0 to use Snowflake, connect your account, and watch as it generates polished dashboards with your data.
Y Combinator
Some of the biggest companies of the next decade won't be software businesses. They'll be services companies like insurance carriers, law firms, and tax practices rebuilt from scratch with AI doing most of the work.
In this episode of Startup School, YC Visiting Partner @CharlieWarren walks through the playbook for building AI native services companies, covering how to pick a market with the right traits, why variance kills these businesses faster than anything else, and the P&L math that’ll transform your business model.
00:00 — Intro to AI Services Companies
01:01 — Picking the Right Market
02:55 — Markets YC Likes Right Now
03:43 — The Sam Altman Test
04:35 — The Right Founding Team
05:28 — Building the Product
06:19 — Variance Is the Existential Problem
07:08 — The Early Demand Trap
07:53 — How to Price AI Services
08:41 — The P&L Walkthrough
09:33 — AI Operating Leverage
10:27 — Don't Buy Your Way In
Adish Jain ☕️
we’re launching Motion, the frontier agent for tasteful motion design.
this launch video is made entirely with Motion. 👇🏽
QT + comment "MOTION" to get 1,000 free credits.
tag @motion_so in any X post for a surprise.
now generally available at motion [dot] so.
Charlie Warren
http://x.com/i/article/2062053874710810624
So proud to say Phoebe Maffei won her election to Superior Court judge
We have one more judge who will stand for rule of law in our city and this is an important win for common sense
Garry's List: She bartended through law school. Spent 15 years prosecuting violent crime, elder abuse, and public corruption in SF. Now she's running for judge.
"Bar to Bar to Bar" by GL Studios. Vote Phoebe Maffei for Judge. June 2nd.
Prop D was defeated even though the hard leftists spent more than $3M to try to stick it to businesses and working and middle class citizens by dropping an 800% gross receipts tax
Voters saw through the fake “overpaid CEO” marketing lies
Garry's List: Politics isn't abstract. The same decisions that cost businesses billions shape what your family pays at the pharmacy, the grocery store, the gas pump.
Every generation feels the choices we make now. Sitting out an election isn't neutral. It's a choice someone else makes for
Reed Schwartz
The map for Scott Wiener last night doesn't really track prog vs. mod, it tracks YIMBY vs. NIMBY. The bulk of precincts that went for Chan were conservative, high-homeownership, low-density areas on the West Side.
🚨 NEW EXPERIMENT 🚨
Dreambeans is a new, experimental mobile app that uses Personal Intelligence to connect to your Google apps. Every day, it delivers collections of personalized stories, surfacing things you might otherwise miss, alongside topics that are relevant to you, to help you dive deeper into the things you care about most.
Available starting today for eligible US-based Google AI Ultra users (+18), with an open waitlist found on our website below!
Learn more at http://labs.google/dreambeans
Replit ⠕
You shipped your app. Now what?
Your app may look great, but if no one can find it, it stays invisible
Publishing is only the beginning
Meet SEO Agent. It runs a scan for you and suggests fixes to help your app get discovered in web & AI search
Kyle Daigle
Chopped it up with @swyx on @latentspacepod and we ran the gamut on this one. We talked platform, how roles are evolving, the agentic era, the future of open source, and what we’re building next. Spoiler: check out the @github Copilot app. 😉
Latent.Space: 🆕GitHub's Agent Era: 14x commits, 200M developers, Copilot’s next act https://latent.space/p/github
@github COO @kdaigle explains why AI agents are forcing GitHub to evolve beyond code hosting, how Copilot is moving from autocomplete to CLI, desktop, cloud agents, and ambient
SEO Agent!
Replit ⠕: You shipped your app. Now what?
Your app may look great, but if no one can find it, it stays invisible
Publishing is only the beginning
Meet SEO Agent. It runs a scan for you and suggests fixes to help your app get discovered in web & AI search
Ankit Gupta
deeply shocking that Bain and McKinsey didn't find value in AI
John Cassidy: First it was MIT and McKinsey. Now Bain finds that returns to corporate AI investments are disappointing.
Amjad Masad
Benchmarks place GPT 5.5 as the best model on SWE, but is it the best at making apps end-to-end?
Turns out Opus 4.8 continues to be the king of vibe coding on both price & performance.
Introducing ViBench: the first benchmark for app creation based on real world tasks
Amjad Masad
Re Paper: https://dl.acm.org/doi/10.1145/3786335.3813162
Website: http://vibench.ai
You can run but you can’t hide from B2B SaaS
Meghan Bobrowsky: New: Meta today launched an AI agent for businesses that can answer customer questions, book appts & close sales
Eventually it will be able to run their entire business, Zuckerberg said during the launch announcement
It's part of Meta's broadening beyond its core ads biz
Paul Graham
It's encouraging how the tech world has rallied round Ron Conway since he disclosed that he'd been diagnosed with cancer. So many people are helping him in so many different ways. It's really making a difference.
Don’t miss it!
tigris: speaking ++ showing emotion alchemy at @Replit vibecon on June 17-18 in NYC! code! tech! art! and spike jonze!
Lenny Rachitsky
We have a winner of the Lenny's Newsletter x @Replit buildathon 🏆
Congratulations @andrey_esipov on your mind-blowing project, Operators.
Y'all need to check this out: https://operators--andreyyesipov.replit.app
Operators is a Street Fighter II-inspired fighting game where every fighter is an operator from Lenny's Podcast. Every special move maps to a framework that the operator talks about on the show, linked to the episode and timestamp it came from.
So much craft and personality went into this project. Our team was genuinely laughing while testing it.
Congrats Andrey!
And congrats to the 100+ of you who entered the contest. This was a tough choice!
A special mention of a few more projects that came in close behind:
+ Little Lenny's https://littlelennys.replit.app
+ Lenny's Diner https://lennysdiner.replit.app
+ LennyOS https://lennyos.replit.app
+ Oracle of Lenny https://lenny-oracle.replit.app
+ Pod-Squad https://pod-squad.replit.app
Thanks for playing! More fun excuses to build with AI to come.
Lenny Rachitsky: Announcing the Lenny's Newsletter x @Replit Buildathon
Build something awesome with my podcast and newsletter data using Replit, and win fabulous prizes:
🔸 1:1 career coaching session with @amasad
🔸 $5,000 in free Replit credits
🔸 A free year of Lenny's Newsletter
This is
Mada Seghete
Re @swyx and I are curating the AI in GTM track at @aiDotEngineer on June 30.
The thing every AI engineer must realize: GTM just became an engineering problem.
Outbound = agent design.
Enrichment = retrieval + data quality.
Attribution = a knowledge-graph problem.
Forecasting = evals.
OpenAI
It's time to fly.
Chris Elmendorf
Sign of YIMBY times: a well-connected candidate for the S.F. Board of Supervisors who ran on opposition to upzoning got trounced 70%-30% in the richest district in the city.
I guess the neighborhood "felt that [she] failed that test."
Sundar Pichai
Our new Gemma 4 12B model hits a sweet spot between size + performance: it can run locally on a laptop, while enabling powerful multi-step reasoning and agentic workflows. Can’t wait to see what the community does with this one!
Demis Hassabis: Celebrating the milestone of a massive 150+ million downloads of Gemma 4 with the release of the new Gemma 4 12B model! It's incredibly powerful for such a small model and it’s tiny enough to run locally on a laptop with just 16GB VRAM. Apache 2.0 license - happy building!
We've put out a blueprint for democratic governance of frontier AI, and how America can build durable institutions for frontier AI safety:
OpenAI Newsroom: There’s real momentum right now for AI safety policy. Yesterday’s EO on cyber was an important step forward.
We’re proposing a set of ideas for policymakers to consider next and to put the US out in front on frontier safety.
https://openai.com/index/frontier-safety-blueprint/
Mark Warner
It seems there is an entirely different set of rules for Republicans and Democrats.
Dereck Paul, MD
The new frontier of AI for clinicians is understanding thin harness/fat skills and embracing agents for practicing medicine.
Dereck Paul, MD: http://x.com/i/article/2062191955589976064
Ruohan Zhang
Excited to introduce StereoPolicy, led by @EvansXuHan.
📷📷🤖StereoPolicy is an effective way to add geometric cues to modern robot policy models while keeping the strengths of pretrained 2D encoders.
⁉️Why stereo for robot manipulation?
Monocular RGB often lacks the depth cues needed for precise manipulation, while RGB-D and point clouds can be noisy or brittle, especially on reflective and transparent objects in real-world deployment.
Instead of explicitly reconstructing disparity, depth, or point clouds, StereoPolicy directly fuses synchronized left/right RGB views to learn implicit stereo cues, avoiding extra reconstruction latency that can make real-time manipulation difficult.
Project Page: https://stereopolicy.github.io
fly with codex
OpenAI: It's time to fly.
If you've got to shard postgres, accept no substitutes
Lev Kokotov: 1M PgDog downloads! 🚀🚀🚀
SF is pro-growth pro-business, and we have to keep winning
Keep voting, keep talking about the real issues, don't get distracted by culture war, keep building
Andrew Jeffery: Pretty clear win for moderates in San Francisco's election yesterday.
Most outsiders think this shift started with the Nov 2024 election of Mayor Lurie.
But this movement started years prior.
From our SF multifamily pitch deck printed in 2023:
"Political winds are shifting
kasra
this is an interesting point in the new ted chiang piece – no one really claims that alphafold is conscious, or that sora or midjourney or dall-e are conscious
Codex is absolutely killing it this afternoon for editing spreadsheets for me.
San Francisco is healing. We don't fall for this type of ridiculous gaslighting anymore.
New York Post: Ex-AOC aide's $10M campaign ends in humiliation as he bombs in every precinct https://trib.al/c2FqPM0
Alec Stapp
This new research on US unicorn startups is really interesting.
Some key facts from the report:
1. Immigrants founded or cofounded 455 of America’s 775 privately held billion-dollar startups, equal to 59% of all US unicorns.
2. 66% of all US unicorns were founded or cofounded by immigrants or the children of immigrants.
3. 79% of US unicorns have either an immigrant founder or an immigrant in a key leadership role.
4. The 455 immigrant-founded US unicorns have a combined valuation of $5 trillion.
5. That $5 trillion valuation is larger than the total stock-market value of companies listed in all but 7 countries.
6. Including immigrant-founded unicorns that went public since 2016 pushes the total value above $5.8 trillion.
7. The number of immigrant-founded US unicorns rose from 50 in 2018 to 455 in 2026.
8. 24% of US unicorns have a founder who first came to America as an international student.
Major upgrade to GPT-Rosalind, with much better intelligence for drug discovery, analysis, design, and experimental workflows:
OpenAI: We’re bringing new capabilities to GPT-Rosalind, a model series purpose-built for life sciences research at enterprise scale.
It brings GPT-5.5’s agentic coding and tool use together with stronger intelligence for drug discovery, analysis, design, and experimental workflows.
Daniel Lurie 丹尼爾·羅偉
To anyone who comes to San Francisco to commit crimes: we will find you, and we will prosecute you. On behalf of San Francisco, we are wishing Officer Brittney Taylor a full and speedy recovery. Our officers have our backs, and as mayor, I have theirs.
San Francisco Chronicle: The 36-year-old man charged with attempted murder for allegedly shooting a San Francisco police officer had been out on parole for less than six months at the time of the Sunday night shootout in the Bayview, according to officials and court records. https://www.sfchronicle.com/crime/article/sf-officer-shooting-suspect-charges-22289904.php?taid=6a208b196ddeab0001696c5a&utm_campaign=trueanthem%2B3988&utm_medium=social&utm_source=twitter
Jesse Genet
I'm so fucking bullish on the United States and so dismayed that the average citizen is not so fucking proud that the best fucking AI companies are US companies
Let's get it together folks 🇺🇸
Sry if you came for the gentle homeschool poasts and got this instead 😅
Jelani Nelson
“In addition to overreliance on AI, Garcia also pointed out that many students are underprepared mathematically, a concern echoed by campus associate teaching professor Gireeja Ranade…
Both Garcia and Ranade have joined more than 1,300 UC faculty in signing a petition calling for the reinstatement of ACT and SAT standardized testing scores for STEM admissions in the UC system. The petition and its accompanying open letter detail similar concerns with students’ mathematical preparation.”
Litong Deng: Students are failing UC Berkeley CS classes at an alarming rate. More than 35% of students failed CS 10, a course described as “a gentle but thorough introduction to computer science.” In the past few semesters, less than 10% of students failed the class.
Latent.Space
🆕Scaling Past Informal AI
https://latent.space/p/axiom
@axiommathai founder & CEO @CarinaLHong explains why math may be the missing path from code agents to AGI, why verified AI is about scaling brilliance not just fixing hallucinations, how Lean and formal proofs turn reasoning into a stronger reward signal, why Axiom sees the TAM as all AI-generated code, what it means to prove research conjectures in a self-verified way, and why the next bottleneck for AI may be not generation but verification.
AJ Asver
At @ParchaHQ we saw coding costs explode so we built our own IDE that does routing at the harness level e.g. Claude Code for planning, Codex for long running execution, Cursor for bugs. Plus custom routing, remote sessions, a browser inspector and voice interface.
Now it's open-source! 👇
x.com
stereopolicy.github.io