Arne Strickmann retweeted
Sam Altman created the most ruthless model in the world
we built the biggest mafia-like game ever
> 250.000 humans playing with/against AI
we found:
@OpenAI → maniac LIARS
@SpaceXSI → by far the most LOYAL
@AnthropicAI → the biggest SNITCH
we also added a flybrain ;)
Arne Strickmann retweeted
Enterprises which win the AGI race will build their own internal Mercor.
Until now there was no way to build internal human datalabs without surveilling your workforce.
Introducing @alexandriainc
Alexandria emerges from stealth
We now force personal agents like @bot, @Muse or @instinct to tell us why they want to make tool calls.
This lets us group and visualize the background work agents perform, which is usually opaque, and see which agent did what, and when.
It's pretty cool to expose this data back to the agent and ask "why did instinct change this document" or "what prompted muse to write this email".
@hkonsti_ 🫡
While personal agents like @instinct, @bot or @Muse are getting really good at long-horizon tasks, there is also an entire class of problems where I need something simple done right now.
We keep building new widgets for these. Long term, they could be generated on the fly with generative UI, like json-render from @ctatedev.
Jev-based and inspired by @anishfn.
Was fun hacking on this with @hkonsti_!
Arne Strickmann retweeted
Today, we're launching Sazabi version 1.0.
The reality of shipping software as a team in the age of AI is actually crazy.
Agents merging their own PRs. Hundreds of deployments per day. A codebase that is constantly changing. Junior developers with 10x or 100x leverage. 😵💫
I spent the last year building a safety harness for teams that ship with AI. It's called @sazabi.
The premise is simple:
1. You send us logs.
2. We detect anomalies, investigate them, and send you back rich, actionable alerts about real problems in your app.
Sazabi helps you find issues before your users notice.
No monitors or noisy alerts. Just pure signal.
And today we're making it available for anyone to use.
Arne Strickmann retweeted
Replying to @ledwards @AstasiaMyers
coffee movement is the actual only S-tier coffee shop in SF, and maybe golden goat. sightglass is basically starbucks at this point
It’s interesting that no one has built an all-in-one tool for personal agents (like @Muse, @instinct, @bot, @interaction) that hold all connections to third party companies. Convinced that this should be an independent product.
That way you can keep auth separate, audit what the agent does, set fine-grained permissions and easily switch between the agent apps.
I built it over the last few days with @hkonsti_. It’s open source and easy to self-host.
Now my Instinct has exactly the same tools as Claude Code on my laptop.
What are we building with K3 today?
Introducing Kimi K3: Open Frontier Intelligence
🔹 2.8 Trillion Parameters, 1 Million Context, Native Multimodal
🔹 Kimi Delta Attention enables up to 6.3x faster decoding in million-token contexts
🔹 Attention Residuals deliver ~25% higher training efficiency at <2% additional cost
🔹 Built for long-horizon agentic coding and self-evolving workflows
Kimi K3 is now live on on Kimi.com, Kimi Work, Kimi Code, and the Kimi API.
Open Weights by July 27, 2026.
🔗 API: platform.kimi.ai
🔗 Tech blog: kimi.com/blog/kimi-k3
Arne Strickmann retweeted
The question isn't whether agents can generate code. It's everything around the model. Where the next category-defining coding companies get built, from @denise_teng25
gradient.com/blog/posts/codi…
Arne Strickmann retweeted
Use Grok 4.5 in Emdash
➡️Grok 4.5 is built for real-world engineering: navigating large codebases, working across multiple repositories, and completing long-running tasks with a variety of tools.
➡️It combines fast, token-efficient reasoning with strong coding performance. It’s served at 80 tokens per second and priced at $6 per million output tokens.
➡️Grok 4.5 ranks #2 on @ProximalHQ’s FrontierSWE leaderboard: behind only Claude Fable 5 and ahead of Opus 4.8, GLM-5.2, and GPT-5.5.
Grok 4.5 ranks #2 on FrontierSWE
It is only outperformed by Claude Fable 5 and ranks higher than Opus 4.8, GPT-5.5 and GLM-5.2
Arne Strickmann retweeted
What's new in Emdash v1.1.38:
- New chat UI powered by ACP
- GPT-5.6 Sol, Muse Spark 11, and Grok 4.5
- Notion integration
- Oh My Pi and Zero CLI support
- Storage management
- Drag terminals into task tabs
- Voice mode support on macOS
- Task archive shortcut: ⇧⌘E
Plus lots of fixes and improvements.






















