How can OpenAI be so simultaneously ahead on math and physics but so far behind grok on consumer bots.
Dot is so bad compared to Grokbot and it costs 3x
My biggest learning for openclaw, Hermes and grokbot is to have a verifier bot that validates actions from other bots to ensure tasks are actually done when a bot think they are done. Doubly true for accounting or reconciliation.
Finally moved to GrokBot after being on Hermes and openclaw before
It is so much better. It’s like going from Arch to Mac.
Austin retweeted
Anthropic released code-modernization v1.0.0, a Claude Code plugin for modernizing legacy codebases
Save this post if you have a legacy project worth modernizing
What I find interesting is that it doesn’t just rewrite code. It tries to prove the new implementation behaves like the original
Here’s how:
- Analyzes the legacy system and extracts business rules
- Generates a modernization plan for your approval
- Rewrites or upgrades the code, even across languages
- Runs the old and new implementations with identical inputs and compares outputs
- Generates additional edge cases to test behavior
- Introduces deliberate bugs to verify tests actually catch them
- Produces a verification report with PROVEN, PARTLY PROVEN, or NOT PROVEN results
It also keeps the original code untouched and requires human approval at critical steps
I think this is the right direction for AI-assisted modernization, generating new code is one thing
Proving that you haven’t broken existing behavior is a completely different challenge
Install:
/plugin install code-modernization@claude-plugins-official
Repo:
github.com/anthropics/claude…
I currently have an AI setup that I would call a mixture of harnesses (hermes, dot, codex, muse). I'm missing something to ultimately glue them together, like a memory and document management system.
Anyone using something that makes always on and ad hoc systems cohesive?
Setting up a math tutor app for kids with varcel and there is no longer a new project button. Everything is in a chat interface in a separate domain.
Is this Peak AI or the way of the future?
Austin retweeted
Apple says you need a Mac to build iPhone apps.
Today I built one on Omarchy Linux, submitted it to the App Store, installed it on my iPhone over USB, and stopped it at a breakpoint in the debugger.
No Mac anywhere.
It's all open source:
github.com/joshuaswarren/oma…
Austin retweeted
I've decided to finally release my vulnerability hunting methodology that I've been using for several months now, both in testing as well as live engagements with great success.
I was inspired by (and huge congrats to) the @Blackfrost_AI team open-sourcing Cyber-Frost Harness, their reproducible security-agent runtime with procedural skills, an evidence ledger, and isolated Docker targets for red/blue/purple work. 🎉
If you like the Cyber Frost Harness, and want some out of the box vulnerabilty hunting and POC development kit to go along with it - check out my VulnHunter.
It's already loadable: their harness reads standard SKILL.md skills. Point skill_dir at VulnHunter's (or run its installer), and the hunting, falsification, and PoC disciplines load immediately.
Repo: github.com/nealbridges/VulnH…
CREDITS: This is fork of @CapitalOne original repo. I simply took it a step further on POC development and model agnostic. I'm also actively developing in my forked repo as I make more changes to it and welcome the contributions from the community as well.
I think I'm doing the whole "respect who you give credit to" drama, and I asked my model to double and triple check it. So i can blame the model if its wrong. Otherwise, dont put me on blast on X if i suck at credits. /s
Looking forward to developing on this further with the community.
Austin retweeted
We'll be spending a lot more time trying to understand the outputs of language models. A few thoughts, tips & tricks:
Writing. Something I've had success with: Ask your LLM to explain something in ASD-STE100, it's a controlled language specification originally developed for aerospace maintenance documentation. LLMs well-versed in this language and it comes with heavy constraints on clean writing style that I often find a lot more readable. Sometimes I've tried to soften it a bit e.g. ask for "80% of the way to ASD-STE100" because the spec is quite stringent. But even better:
Diagrams / images. Instead of writing, ask your LLM to create a diagram. These can be a lot easier to process, parse, and understand. But even better:
Web pages. Ask for output "in HTML" to get a beautiful, interactive webpage. LLMs are getting really good at frontend and can create beautiful experiences, animations, etc. But even better:
Explainer videos. The output format I am most bullish on is fully custom / bespoke explainer videos generated on any arbitrary topic. Experiment with things like "Create a 3b1b style video explainer on X. Use my ElevenLabs API key for audio narration". (you'd need an API key for the latter or you can ask your LLM to find you decent free alternatives that use your local compute). This is actually starting to work!
In summary:
- As LLMs get better, they will do more and more of the legwork autonomously, and a lot more of our work will rise up the abstractions into oversight and understanding.
- Luckily, LLMs can help here too because as intelligence and code are increasingly abundant, you can ask for large, custom, discardable software artifacts (e.g. web apps, video explainers) that would have never made sense to create before. Push the boundaries here and you'll be surprised.
Austin retweeted
"Let’s plan the roadmap for next quarter” is the modern day equivalent of “let’s fax everyone the strategy.”
We're going into q4 planning and it makes my brain hurt. Roadmaps existed because software was expensive to build. You needed a bunch of people. Meetings. Someone moving little boxes around Jira for three months from now. It makes no sense and I'm trying to work out what we call the replacement. Maybe just "Continuous Compilation", ideas just compile to working product. I want a list of ideas that we want to continuously compile. That's the roadmap now.
The cheapest way to figure out whether something is a good idea is often to just build the fucking thing.
Here’s how I work now. I've done this weekly over the last few months and it works. My current napkin math is that I can usually build 6 months of old-school roadmap in a week. Sometimes more. I've also hit a wall sometimes. It happens.
When we have an idea, I usually 10x the ambition and then write a prompt, not more than a page long, that basically says:
- here’s the ambitious goal
- here are a few use cases
- here’s roughly how I think we should build it in terms of where it fits into code base, design considerations, etc..
- here’s how we’ll know if it actually works
- now go fucking surprise me
Agents build all night. Not even with a crazy factory or complex agent or agents. The base models are just too good for that now.
In the morning I look at it. Ok, they usually build in a few hours. Usually some of it is great. Some of it is stupid. Some of it exposes that my original idea was wrong.
And almost every time it gives me 5 new ideas I couldn’t have planned upfront. So we throw shit out. Who cares? We learned something.
Then we change direction and send them off again. It usually takes 3 rounds before the shape appears and we've learned about the gremlins and possibilities.
95% of the time I end up building something with 10x the scope of what somebody had carefully spread across the next quarter in Jira.
“What should we put on the roadmap for next quarter?”
My brain now translates that to: “Oh, you're stuck in the 1500s”
Austin retweeted
🚨 IMPORTANT ANNOUNCEMENT TO THE RUBY COMMUNITY 🚨
_WHY'S SHOES LIVES
BUILD A RUST-GRADE DESKTOP APP IN RUBY TODAY
RUBY THINKS. RUST DRAWS.
cc @cooperx86 for any media enquiries
1/6
Amen. Right people in the right seats.
GET THE RIGHT GUYS ON THE BUS!
"Mediocre people don’t like high achievers and high achievers don’t like mediocre people ... Get the right guys on the bus, get them in the right seats, and get the wrong guys off the bus."
~ Nick Saban video from @TheCrimsonWhite
I don't understand how google is so far behind the 8 ball on personal assistants.
They have a lot of the world on gmail, a lot on google drive. You wouldn't have to give another provider more data.
Openclaw was over half a year ago - Seems like a huge opportunity miss.
anthropic.com/research/glm-5…
Want regulatory capture so bad they can taste it
Austin retweeted
🔒 HotCell moves risky Active Storage processing out of the Rails application and into isolated containers with no network or credentials, limiting the damage from vulnerabilities in tools such as libvips, ImageMagick, and ffmpeg. #Rails github.com/basecamp/hotcell
Austin retweeted
Campfire compiled with Roundhouse + Spinel runs 5.4 times faster in rps, and consumes 1/12 of memory. No modification needed.