Our video transcoder at @gumroad runs on Hetzner, so every video pulled from S3 was billed as internet egress: $935 last month. Now it pulls via a signed CloudFront URL. S3 to CloudFront is free, and our private CloudFront pricing is far cheaper. ~$860/month saved, ~$10k/year.

Today, we were alerted of malicious activity that compromised Railway's servers We are sharing this with you broadly and swiftly. We recommend folks rotate their keys out of an abundance of caution.

I’m only using dots for everything I used to do on Chat and Work 🤯

yeah u didn't understand my point LoC being moat is dead, copying something is much easier than before... that's what I mean by software is dead

my new job is member-of-press-tab-and-return-staff @ openai no but actually i am spooked at how accurate these recommendations are

Today, we're releasing Claude Code for marketing. It does a marketer's work in minutes by browsing, clicking, and posting like a human would. The marketing hire is now optional:

🫠

Proactivity is one of the defining features of Dots. I think Grok Bot is super cool and being built by a legendary team of builders. I’m glad they were able to add proactivity to their product as well! @poteto we can work off of a shared roadmap, delegate to dots and bots, go grab some beers, and see what happens :)

Now that I’ve safely landed, I can say I’ve taken my first flight with dots :) gob checked me in, got my boarding pass and helped fill out Singapore’s arrival form. Then handed me its cloud computer for the final declaration. That bit’s on me to legally declare, of course. Had to ask this time. It’ll notice this habit and remind me, before I’m awkwardly standing at immigration 😂

Pretty excited for DevDay tomorrow. We have found a new thing.

Tons of folks ask me "how do I create a CODING_STANDARDS.md file?" My answer is that if you're using it right, it should only be empty for about 5 minutes. Every meaningful thing your agent gets wrong should be noted into CODING_STANDARDS. - Forget to implement dark mode? - Forgot a validation rule? - Used the wrong framework primitive? Use /retro to fill it up, or modify it yourself.

This is great but can we have shorter clips and not 30-45 mins videos, just giving out conclusions and crux? Its difficult to watch an hour long video everyday morning

We just ran Jev on our WebMCP benchmark. The result: basically broke the benchmark. Jev + Mercury 2.5 (a fast, low-cost LLM) using WebMCP solved 100% of the tasks at roughly 112× lower model cost than GPT-6 Astra using computer use with code execution. Compared to Astra using screenshot-based computer use, the model cost was 245× lower (!). We also compared Jev operating the browser with and without WebMCP. We used Browser Use’s open-source Ultrafast, with some improvements to the harness to make it more reliable across the benchmark. Jev’s browser-control accuracy on its own was not amazing - adding WebMCP nearly doubled the number of solved tasks, from 25/49 to 49/49, while reducing model cost by 18% (more on why below). The benchmark and methodology are fully open and reproducible. Full results: webmcp.com/benchmark A few words on how the Jev + WebMCP harness works and why this is exciting: Jev receives text as input and a set of discrete options it can choose from. With WebMCP, those options are the tools exposed by the website. At each step, Jev sees the task, the available tools and previous results, then picks what to do next. The limitation is that Jev can’t generate arbitrary text, which you need for tool arguments. For example, it can choose the search_products tool, but it can’t generate the search query itself. So we split the work: Jev picks the tool and Mercury 2.5 generates the arguments if needed. This works well because turns out most of the cognitive load in these tasks is around choosing the right action. The argument generation itself is relatively simple, so we can delegate to a small and very fast model. We used Mercury, which outputs 1,000+ tokens/sec and is very cheap. The result is a pretty simple combination: Jev for tool selection + Mercury for arguments + WebMCP for the interface. It ends up being very reliable, very fast, and very cheap. A few words about Ultrafast and why do we think it underperforms: Without WebMCP, Jev chooses from the page’s controls: which button to click, which field to fill, or which option to select. But choosing a valid button is different from choosing the right next step. The agent still has to navigate menus, understand forms, recover from errors and recognize when the task is actually complete. Our hypothesis is that WebMCP makes the decision space much simpler. Instead of figuring out a sequence of clicks through a website, Jev chooses explicit actions that directly advance the task. @typesafeai itself documents weaker accuracy on questions requiring multiple reasoning steps. WebMCP moves much of that complexity into the website’s tools, leaving Jev with clearer decisions and fewer opportunities to go wrong (in a sense WebMCP "compresses" a sequence of clicks into one tool call). Our modified Ultrafast setup solved 25/49 tasks - that is a result for our particular implementation and benchmark, not a universal limit on Jev or Browser Use. We are open to more harness optimization to get this result to perform better, feel free to directly contribute to the benchmark here: github.com/nekuda-ai/WindTun… Browser-use ultrafast: github.com/browser-use/jev-u…

we just reset everyone’s muse token usage. enjoy!!!