One fun thing about building stuff with AI is how you can make really amazing internal tooling.
This is a @gbrainio LiveView tool where I can perform manual upgrades to brains or follow along as an agent calls the same upgrade tooling to perform machine upgrades.
Elixir v1.20 released! Now officially a gradually typed language: Elixir type checks every single line of code, finding bugs and dead code, without developer overhead (no typing signatures) and extremely low false positives rate. Plus a faster compiler! Links and reports below.
GBrain is free MIT-licensed open source and available at https://t.co/yFpFU4pn5b
I use it with my own 350k markdown knowledge LLM wiki + OpenClaw/Hermes Agent setup. You can have my setup in about 30 minutes. (The agent does most of the work!)
@sama Hey @sama, love ChatGPT and all of the progress. Could you guys look at multiple account support? It's a PITA to switch between my enterprise and personal account all the time.
I'm using it to build @gbrainio. You can't really tell from this screenshot, but Elixir is connected to a bunch of different servers coordinating agents, servers, & UI seamlessly. It's bonkers.
When scrolling down, LiveView debounces the "recent message" cursor on the server.
Tokenmaxing some changes for the core @gbrainio harness tonight that I hope fill make for a sane OpenClaw.
The idea is to give people the raw power of OpenClaw & gbrain while making it easy to recover from oopsies.
Like a souped up Lambo with a great insurance policy.
🏎️💨
Early review of GBrain - I am working on it day-to-day, dropped more than 20 PRs in the last week, so it is a young codebase growing quickly and changing. This is great feedback.
GBrain getting better every day
https://t.co/S2r0Aifw4v
Q: How are job postings for software engineers rising rapidly despite AI agents automating coding?
A: Because there’s far more code to manage than ever before. We’re already seeing a 14x YoY increase in GitHub commits, and it’s accelerating.
AI has dramatically lowered the cost of writing code, so it’s now being used across far more businesses, applications, and use cases.
We’re at the beginning of a massive productivity boom driven by the proliferation of bespoke software throughout the entire economy.
Coding has been AI’s breakout use case this year. The fact that it’s increased demand for software engineers — rather than decreased it — should call into question the entire “AI will cause mass job loss” narrative.
"let it cook" is the line Anthropic engineers just repeated all day at Code with Claude London.
Boris Cherny said it in the keynote. Ravi Trivedi said it in the next talk. Katelyn Lesse said it at the panel.
it means: stop micromanaging the prompts. write the routine. let Claude prompt itself.
his framing:
> routines are higher-order prompts.
> the runtime is shipped.
> the prompts are the bottleneck.
what they didn't say on stage: most routines die without 3 specific properties.
i tested 30. 9 made it. the other 21 violated one of the three.
the 9 that survived are in the article.
all verbatim, just copy-paste.
worth more than $300 of "prompt engineering consulting" before you build anything.
At some level you just have to embrace it. If it looks good, who cares?
BTW my images are all AI generated too.
Is it perfect? Far from it, but when devs run into something that looks crappy, they spend a bit of extra time prompting something better.
What’s new is old again.
The other thing you need are strong guard rails.
I’m written in Elixir, so there’s a bunch of tests, test coverage analysis, module boundary enforcement, and linting to keep things in the rails.
The current 1400 test suite runs in about 2s.
When your agent tells you to look at or do something on its behalf to figure out why its not working, close the loop by giving it the the tools it needs to look at it and fix it itself.
This is a core skill you have to learn to build stuff with AI.
That and strong guardrails.
@sflorimm There’s going to be more companies that are just markdown files.
But if you really think about it, hows that different than any other SaaS? Code is text files. Markdown is text files.
It’s text files all the way down.