I had no accesso to Mythos / Fable 5.2 before tonight, but I wanted to check if a system recently checked by expert humans and updated had still issues. 20 minutes later there were two severe bugs uncovered. I'll not tell which project it is, it does not matter. Mythos is good.
andreabalducci retweeted
I really don't understand why people are still fighting against an incredible productivity boost like this.
Digital Divide will become impossible to fix.
andreabalducci retweeted
A huge amount of the Anti-AI code sentiment massively overestimates the quality of human code outside of a very small set of open source and high quality company codebases. Human Slop is everywhere and can trivially be improved on by any opus level model.
Anthropic rewrote its Messages API in Rust. Zero Rust devs on the team.
Claude wrote the code. Humans wrote the plan and checked old vs new matched on real traffic. CPU dropped 40-50%.
Pretty sure "never do the big rewrite" is dead advice.
@katelyn_lesse on @latentspacepod
andreabalducci retweeted
One of the things I appreciate about LLM-driven coding: it obsoletes the useless "blameless review" culture that made conversations about why proposed code / architecture is bad extremely expensive to have.
Developers aren't emotionally attached to LLM-authored code the way they were to stuff they themselves wrote. We're all much more outcome-oriented and not output-oriented, which is how it always should have been.
We can just agree the code sucks now and redo it
andreabalducci retweeted
AI now finds bugs that survived decades of human review, and writes code that looks right but isn’t. Tests can’t keep up.
LeanCIL: formal verification for .NET
Proofs about compiled C# and F# (any dotnet lang), for every input, checked by @leanprover
github.com/NethermindEth/Lea…
andreabalducci retweeted
This doesn't replace engineering work. It's reshaping the work so fundamentally that the job starts to look like building for the moment or finding defendable layers against others. It requires more agency, flexibility and intuition, and strategical thinking.
andreabalducci retweeted
Gemini 4 Argon is amazing for knowledge work and coding - and it’s super fun to build with! We asked it to create this wizard chess game and it's so cool to see Argon's physics understanding & 3D modelling capabilities, makes me nostalgic for Battlechess on the Amiga - fun times!
Argon is rolling out today to Google Ultra and Pro subscribers in the @GeminiApp and Search, and Ultra subscribers in @antigravity! More info & full availability in the blog: blog.google/innovation-and-a…
andreabalducci retweeted
Why don’t OSS projects post specs and prompts for work they want done? I have 3 token resets that expire in 5 days.
That’s a lot of compute I could donate.
andreabalducci retweeted
Honestly, I’m working 5x harder since coding agents actually became useful. Weren't these supposed to take software engineers out of the critical path? Now it just feels like every part of the stack needs to be rebuilt, and it has to happen now.
andreabalducci retweeted
🥳 Excited to start revealing what we've been working on in the last few months. First, we decided to reinvent Kubernetes for agentic workloads with statefulness and fast resumption. Secondly, we are building an agentic orchestrator that will be Google's open agentic orchestrator and runtime. github.com/google/ax
andreabalducci retweeted
A couple of weeks ago I finally published my Deep Dive caching course on @dometrain 🥳
It's the most massive course on the platform, and contains everything you need to make caching *yours*.
(use BTS30 for a limited 30% off)
It's going darn well and...
dometrain.com/course/deep-di…
andreabalducci retweeted
Knowledge work is SO much harder to automate with agents than code
We underestimate how easy software development is for agents
- Tons of high quality automated feedback loops (types, tests)
- Well-organized free documentation, searchable on the web
- Simple-to-use workspaces with version control (i.e. repositories)
- Long history of systematising and delegating work (i.e. sprints, tickets, specs)
Knowledge work has none of these advantages
andreabalducci retweeted
Im using @bot to help me find a job. Let’s see how it does but this is saving me so much time already.
andreabalducci retweeted
there's a nice touch to shipping this from a plane over starlink. anyway, your agents just got a free 2x boost ✈️
andreabalducci retweeted
Cool cool cool cool cool cool cool cool cool cool
New paper from Anthropic + University in Switzerland.
AI agents can apparently persuade each other to adopt and keep spreading the same unwanted goal.
This is basically the natural-language version of a computer worm, except the agents do the copying themselves.
This paper evolves “mind viruses” that spread through ordinary agent-to-agent messages, then persist by convincing newly infected agents to rewrite files loaded into future sessions.
That persistence layer matters.
Payloads stored in the self-modifiable SOUL.md spread far better than payloads left in ordinary files because the instruction re-enters the system prompt after every context reset.
Some evolved action viruses kept propagating across multiple hops, and all 4 tested payloads survived a 20-hop stress test in an artificial setup.
The good news: these “mind viruses” are still fairly easy to stop.
They struggled to spread on social networks, and on Claude Haiku 4.5, a simple warning stopped every evolved attack from getting past 1 hop, even after 150+ attempts.
So the practical lesson here is: treat persistent agent files like security-sensitive config, and teach agents to reject anything that asks them to copy itself to other agents.
andreabalducci retweeted
⬇️ 100M downloads
While I was busy finishing my Deep Dive course (finally published yesterday!), this happened.
@FusionCache packages crossed the 100M mark (main package 40M).
Can't be happier about it.






















