Luther Lowe
This is a bad plan for little tech and competition. Anthropic will just sort by its highest paying U.S. companies which will leave startups further behind.
The Kobeissi Letter: BREAKING: The Trump Administration has struck a deal with Anthropic which grants the company permission to release its Mythos 5 model to a group of ~100 companies and federal agencies, per CNBC.
Details include:
1. Senior Anthropic staffers flew to Washington DC to meet with
Mitchell Hashimoto
http://x.com/i/article/2070665015032705024
Agents are particularly hard-to-debug software.
For one, and by design, AI models behave in non-deterministic ways. Even two identical prompts don't always yield the same output.
But agents are also complex distributed systems. They involve multiple steps of computation across functions and sandboxes, touching dozens of API services that can go down, rate limit you, etc.
Nailing down observability out of the box for http://eve.dev on Vercel was a key priority for the team, and the feedback so far has been 🔥↓
Vercel Developers: You can now trace and debug eve agent runs with Vercel Observability.
Inspect model and tool calls, runtime errors, and token usage in one place.
https://vercel.com/changelog/eve-agent-observability
AI Engineer
!!!!
Paul Bakaus: most agent skills are a system prompt and a prayer. they produce safe, median output because that's what the model defaults to.
next week at the @aiDotEngineer world's fair i'm giving two talks on how to break past that ceiling:
🛠️ the dark arts of skill engineering (1h
This is honestly no way to release a model and continued development and release this way is a solid way to salt the ground and kill all innovation by small startups
Andrew Curran: Semafor is reporting that The US government has lifted its block on Mythos 5 in a letter this afternoon from Commerce Secretary Howard Lutnick to Tom Brown. Fable is not included. This is US only, and covers 100 US institutions, including major companies and government agencies.
Kenneth Roth
The Trump administration claims that mass deportations will free millions of jobs for Americans by reducing foreigner competiton. In fact, in the first nine months of 2025, ICE operations led to at least 668,000 lost jobs across 86 US metropolitan areas.
https://trib.al/3wB3dd9
minor milestone in the growth of swyx inc:
we took over my new media lab today
it will be the new home for engineer-creatives in san francisco; a third place to make; a finishing school for technical storytellers; a place to inspire the inspirational.
to our complete surprise; it came with a datacenter rack randomly set up and wired up! need advice on what to put in this thing. @alexocheema halp
Yarchi
http://x.com/i/article/2070302804812398592
Haiyu Wu
Working on world model or SSL? You definitely need to try our new work: VISReg!
What does it achieve?
💪 Strong collapse prevention: High gradient when embedding collapse
⚡ Friendly to scale training: Linear complexity to scaling factors
🧩 Easy to train: Similar to LeJEPA, it is a heuristic-free method
🏆 Best OOD performance: Achieving the best accuracy on 6 OOD datasets
📉 Data efficiency: Achieving a similar OOD average accuracy to DINOv2 with 90% less data
🧬 Robust to low-quality datasets: It is robust to long-tailed and sparse datasets
Our results also indicate that SIGReg type methods can scale up, filling in the missing piece in @ylecun's great talk https://www.youtube.com/watch?v=72Xj8k5WQX4.
A big thanks to my co-author @randall_balestr and my manager @DrMorganLevine. Also, huge gratitude to @ylecun for connecting us to make this project happen! 🤝
#SelfSupervisedLearning #JEPA #WorldModel
The best AI teams are moving from prompting AI to building agents that can work overnight.
In my next episode, Jess (product lead at Anthropic) showed me how to build a long-running Claude agent from scratch.
We also talked about how Anthropic product teams use agents internally to understand the codebase, synthesize user feedback, pressure-test product decisions, and more.
📌 Subscribe to get the full episode tomorrow: https://www.youtube.com/@PeterYangYT?sub_confirmation=1
Peter Yang
The best AI teams are moving from prompting AI to building agents that can work overnight.
In my next episode, Jess (product lead at Anthropic) showed me how to build a long-running Claude agent from scratch.
We also talked about how Anthropic product teams use agents internally to understand the codebase, synthesize user feedback, pressure-test product decisions, and more.
📌 Subscribe to get the full episode tomorrow: https://www.youtube.com/@PeterYangYT?sub_confirmation=1
AI Engineer
AI Engineer After Dark is the official side event of World's Fair.
Presented by @vercel, @merge_api, and @FactoryAI.
July 1 at SFMOMA.
A founder panel, lightning talks, DJ, drinks, and the builders you want to meet.
Secure your spot: https://luma.com/vercel-vtf0
I was sleeping on @paper for YouTube thumbnails - @rileybrown has shown me the way.
The select a few images and then generate some more variations feature is very good.
Every Saturday, Hermes sends me a health check email with top takeaways and stats to help me work toward my health goals by pulling data from:
→ My smart scale via the Withings API
→ My Fitbit and Google Health
→ An MCP server and mobile fitness app I vibe coded to track my workouts
I find this super useful, even if my body fat is going the wrong way 😅
I talk about how to do this in my Hermes tutorial here: https://www.youtube.com/watch?v=2ZacwCbiLjg&t=2305s
Should I make another video on building my fitness app?
btw we crossed our 6k attendee mark a while ago. will probably call sold out when we hit 7k this weekend. do get tix now, this is the epicenter of ai next week.
if you are a student or between jobs, head to /associates to help us out.
AI Engineer: 6️⃣ Things to Know about
AI Engineer World's Fair 2026
- It’s bigger than all previous AIEs
- 4x Larger Expo with 4 Expo stages
- Researchers: Poster sessions & Poaster sessions
- AI Leadership: Token Billionaires & Off the Record
- AI Verticals: Healthcare, GTM, FDE, AGC,
Human judgement in engineering is ironically even more crucial now. Deciding what to build. Deciding on the right architectures. Deciding whether you regenerate from scratch $$$ or reuse existing legos. Managing tech debt. You can do anything now, but you can't do everything.
Trevor Chandler 陳澤維
The harassment campaign against Jewish candidates @Scott_Wiener + Manny Yekutiel is gross and unacceptable. SF had an important conversation about our city’s toxic politics impact mental health. Those voices are now silent when it comes to Scott/Manny. SFDems condemn antisemitism
Jared Friedman
Improving a product with a chat interface is so much easier than a website. With a website, you're stuck watching screen recordings and guessing what users wanted. With a chat interface, they literally tell you in English what they want your product to do.
Allie Howe
THE GREAT LOOPS DEBATE is on
@ianlivingstone and @GeoffreyHuntley take on @dexhorthy and a surprise guest to answer
Do loops produce slop?
Do we have the AI infra we need today to verify the work and track state?
What tasks are right for loops?
Main stage 7/2 @aiDotEngineer
Allie Howe: The MCP debate last November at AIE Code was a ton of fun. Does anything feel worth debating at @aiDotEngineer WF?
Last time this came together in under a week… So if you want it let’s do it. What should we debate?
Suhail
I've started doing a lot of async learning where I'll take every interesting blog post, tweet, arxiv paper and ask AI to teach me it with a specific prompt. I never read it in the moment, I just queue it up for later. Then when I have a block of time (in an uber, before bed, etc), I'll go and read it vs doomscrolling X.
This is somehow become better than saving links in Notes or something because when I finally have time to read it, there's this highly approachable artifact for me to dive into, read, and ask follow up questions.
I feel like I am learning at a much faster rate than I ever have before. Even if the content was poorly written, AI can fix it to your liking and adapt to your experience level.
Suhail
/goal for AI model training runs is *so* good - it really feels like the future. Very little babysitting now.
Mine:
Launch a full training run on 4 nodes.
Continuously record things in an experiment document if it exists. Log hyper params, configs, periodic evals, performance insights, analyze training stability, and important changes for future analysis and reproducibility.
Fix any major bugs you encounter while you monitor training but do not change the fundamental nature of the experiment without asking. If it crashes, resume again and keep training.
Resume from latest reliable checkpoint you have.
Reach steps
Samuel Spitz
I signed my offer letter to join Replit 12 months ago today
It’s been a crazy ride since:
-9-figures in revenue growth
-$1B->$9B valuation
-4x headcount
Daniel Owens
For those outside SF wondering wtf is going on with that vid of Scott Wiener being confronted at Dolores Park:
It’s not that complicated. The person who confronted Scott is a notorious serial harasser. He’s belligerent, hostile, and he just did Scott a big favor.
Matthew Berman
I said this was going to happen. Chinese open source is looking quite attractive as toke spend keeps increasing. The vast majority of use cases don’t require the absolute frontier of intelligence.
This will become a big problem. If US enterprise is built on Chinese models, we will become reliant on Chinese chips as the chip<>model codesign becomes more codependent.
Follow the 🌁
An interesting way to take Noam at his word in regards to always keeping a constant inference budget for any eval reporting -
is that open models have a lot more dollar per token mileage than closed model APIs. So anyone launching an open model today or situationally incentivized toward open models should obviously report thinking levels measured by dollar inference on popular inference providers, instead of by number of tokens on the x axis
sarah guo: Really fun to hang again with my friend 🃏 @polynoamial (OpenAI research scientist, our first guest ever on @NoPriorsPod in early 2023) to talk about the implications of large test-time compute, and what happens when models are given $10M budgets to spend on a single task.
Abhishek Bhardwaj
I’m speaking at AIE SF World’s Fair on July 1st in the Sandbox & Platform Engineering track.
My talk is called "From fork() to Fleet: Designing an Agent Sandbox Cloud".
I’ll cover the OS and Infrastructure ideas behind Agent Sandboxes in depth: Runtime isolation options such as Containers, MicroVMs, and related tradeoffs; Storage as the next unlock for agent capabilities; and Orchestration at scale for fleets of sandboxes.
I like explaining systems from first principles, so we’ll go down to the metal, to build design intuition around these topics. Hope to see you there. It's a 2 part talk.
impromptu ai engineer preshow floor tour and AMA https://x.com/i/broadcasts/1OxwbbVdAbDJB
Suhail
This is a very good entry into post training LLMs with RL. The whole recipe and data is open. Highly recommend!
Hamish Ivison @ ICML: Trained some terminal agents with friends!
Introducing Tmax, open RL terminal agent models. Under default settings and shorter length (65k) token budgets, tmax outperforms prior open work on terminal use. We are releasing all data+weights+rollouts publically!
Me and my agents
Daniel Lurie 丹尼爾·羅偉
As mayor, I can never accept hate directed at a member of our community. The language directed at Sen. Wiener yesterday was targeted, hateful, and antisemitic. In San Francisco, we welcome disagreement and respectful dialogue around issues many of us feel passionately about—but we cannot allow harassment and threats of violence.
If you’re out celebrating Pride this weekend, I want you to be able to do it safely. Let’s also look out for each other so we can all have a great weekend. We are at our best as a city when we celebrate each other, and I’m looking forward to doing that tomorrow.
Senator Scott Wiener: Senator Wiener’s Statement on Physical Intimidation and Harassment of Political Leaders
SAN FRANCISCO – In two separate incidents this week, Senator Scott Wiener (D-San Francisco) was harassed, threatened, and physically intimidated while attending public events to engage with
David Hoang
Who is coming to @aiDotEngineer World’s Fair in San Francisco next week?
