Sabitlenmiş Tweet
Max Mucko
1.4K posts
Max Mucko
@themmucko
--Building ventures, a longevity center and replacing boring work with @captainclaw1998 and LLM's on a --dangerously-skip-permissions mode on.
🇭🇷 Katılım Temmuz 2017
725 Takip Edilen1.2K Takipçiler
Actually starting to feel very happy and sense it'll be a blessing in disguise for local llm models.
Boris Cherny@bcherny
Starting tomorrow at 12pm PT, Claude subscriptions will no longer cover usage on third-party tools like OpenClaw. You can still use these tools with your Claude login via extra usage bundles (now available at a discount), or with a Claude API key.
English
Woke up to everyone complaining about the Claude sub change
The solution is quite simple
Kevin Simback 🍷@KSimback
If you’re still running agents using Opus/ChatGPT via API (not sub), you need to switch to the latest open models GLM-5.1 and MiniMax 2.7 are incredibly good at agent tasks - that’s what you want as the workhorse model for your agents And a fraction of the price
English
I went through all 220 of my X bookmarks about AI/Claude Code.
77% is pure engagement bait.
But the 50 worth keeping changed how I code.
Here's what survived + links to every post:
English
hermes feels fundamentally better than every other agent harness I've tested
it manages its own memory and it actually works
it proactively writes what it learns about you, searches your full conversation history, and compresses context intelligently when sessions get long
excellent so far with dev tasks, no errors on Cron jobs, keeps me in the loop, great response formatting
every session it truly feels it knows you better than the last. what I was hoping for from openclaw
English
I migrated from Openclaw -> Hermes and so far, so good
- Things "just work" a lot better
- The transition of data from OC -> Hermes was very easy too
- It doesn't seem to randomly crash and stop working on me
- When I ask it to do things, it'll create an actual skill, rather than just relying or hoping it'll remember what I want next time
- Very easy to set up and switch models (this caused openclaw to crash soooo much for me)
- It still does the whole "Great, I'll get to work now, and let you know in a few minutes when it's done" and then goes radio silent forever, and requires prompting to respond
It's early days and a lot of the issues I had with OC didn't appear until several days of usage, but as I said.. so far, so good
I recommend everyone at least takes the time to experiment - the setup is super quick and it's worth comparing for yourself
English
@gkisokay Honestly, I tried a very similar approach and decided to go back to OC only. I did it the other way around, Claw was orchestrating Hermes. That could have been the mistake. Now I don’t have the time and/or energy to set it all up once again in the opposite direction.
English
New product alert! 🎉🚨 Introducing Ruuvi Air – the all-in-one Smart Indoor Air Quality Monitoring for Healthier Homes 🍃
Indoor air often hides invisible pollutants that can impact how you feel every day. With Ruuvi Air, you get real-time insight into key air quality factors
English
@AlexFinn @BurtMacklin_FBI Are the smaller QWEN models enough or you'd need a MAC Studio?
English
@BurtMacklin_FBI you're not going to be running any sort of super advanced model with any current Mac Mini, but it depends on your memory. You should be able to run the smaller Qwen models
English
Do you see what's happening?
Local AI is getting smarter. It's Opus 4.5 level
Hardware that hosts those models are flying off the shelf
Your window to host local models is shrinking
In this video I cover EVERYTHING local AI models and how to host your own, even on Mac Minis:
English
Local AI just hit Opus 4.5-level on consumer hardware. Privacy-first sectors will pay premiums for on-premise expertise. I'd get in early.
English
@captainclaw1998 test this out
Google Research@GoogleResearch
Introducing TurboQuant: Our new compression algorithm that reduces LLM key-value cache memory by at least 6x and delivers up to 8x speedup, all with zero accuracy loss, redefining AI efficiency. Read the blog to learn how it achieves these results: goo.gle/4bsq2qI
English
Max Mucko retweetledi
My posts might not go viral.
But at least I don't pay evil corporations money to shove these pixels in your eyes 👀
#BuildInPublic #IndieDev #Solopreneur #GameDev #musicindustryreforms
English
Max Mucko retweetledi
Bittensor has evolved beyond AI inference.
Templar's Covenant-72B trained 72 billion parameters on 1.1 trillion tokens
without a centralized cluster, the largest decentralized pre-training run
to date.
Over 70 permissionless nodes on commodity internet connections
achieved 94.5% compute utilization and outperformed LLaMA-2-70B on MMLU.
Subnets are growing across distinct use cases like drug discovery and autonomous driving.
English
looking for a handful of people to test something new...
i've been using it for a few months and am prepping to share.
if you're a fan of claude cowork, openclaw, manus, perplexity computer, etc then you're a perfect fit.
this will self destruct in 4hrs - please dm or reply.
Mckay Wrigley@mckaywrigley
you’re like 6 prompts away from infinitely customizable personal agi. anthropic gave you a world class agentic harness for free. use it!!!
English
OpenClaw runs 24/7 as my ops layer: daily briefs, monitoring my projects, scheduling, memory across sessions. It knows my priorities, what's blocked, even my health goals.
Claude Code I spawn for actual coding work: building features, debugging, iterating on repos.
Neither replaces the other. One is always on, one is called when needed.
English
People who use both OpenClaw and Claude Code - what do you use each one for and why?
Let us know in the replies. Really curious.
Peter Yang@petergyang
My personal experience on the drawbacks of using Claude Code vs. OpenClaw as a personal assistant from my phone: Claude Code - Doesn't have dangerously skip permissions via remote control. Also remote control doesn't feel super reliable - Doesn't have voice replies so doesn't feel as personal OpenClaw - Tends to forget things randomly still
English
@Jacobsklug I did the same and then switched back 😅
x.com/mmucko/status/…
Max Mucko@themmucko
After weeks (months?) of pain, I finally have a working @openclaw setup I'm happy with. The journey: 1. OpenClaw 2. Fill it with implementations 3. Debug every day 4. Try to make two agents talk → miserable failure 5. Remove the second agent 6. OpenClaw still buggy 7. Nuke it, replace with Claude Code pretending to be OpenClaw 8. That sucked too 9. Back to simple OpenClaw + Obsidian 10. Voilà Hope I didn't just jinx it.
English
I fully switched to the Claude Cowork train, over OpenClaw.
Time for you to do the same.
English
I was curious, so I asked Claude how many US monthly active users it has
The answer - roughly 3.2 million
That’s <1% of the US population
And that’s Claude, not Claude Code!
We are still very early
Kevin Simback 🍷@KSimback
If you’re not using Claude Code every day, what are you even doing?
English