@chainlank @peter_szilagyi Tools. You build a tool which provides an api for all of the required signing and then let the LLM call the tool.
English
Timo Reichert
43 posts
Timo Reichert
@thewolfonchain
Founder & Software Engineer | Blockchain, AI & Gaming
Germany Katılım Aralık 2018
217 Takip Edilen2.6K Takipçiler
@thewolfonchain @peter_szilagyi how would they perform any action without keys or access to keys so they can at least receive signed txn outputs from the tool
English
My hot take: an LLM based AI will *never* *ever* be able to keep something it knows, a secret. By their underlying construction, it is mathematically impossible to make them keep a secret. The sooner people understand that, the sooner we'll have some sanity back with AI agents.
owockai@owocki
welp, it happened. @owockibot's hot wallet private key was compromised after only 5 days alive. luckily, funds are SAFE. @owockibot's treasury / signing keys are stored in a safe that requires me to sign. what happened? it was given these instructions to never share it... but it still did! from my investigations, it put the key into git commits (which it swears it didnt push!), vercel env variables (which it swears it doesnt remember doing!), and it looks like it got social engineering attacked through X and telegram (though it swears it didnt share secrets w attacker!). what did i learn? 1. investigations with an agent mediating are hard. i've found my agent is not a *reliable narrator* during the investigations. sometimes it forgets things, contradicts itself - esp between context windows. it may even be covering for itself, i cant know for sure. 2. if you expose your agent to the internet, and give it secrets, you cannot be 100% sure it wont leak them. 3. its still gonna be useful for @owockibot to do small txns itself. i am going to be coming up with a way to let it do txns via @MetaMask UI for now. but if anyone is building an agentic wallet hit me up i want to try it out. 4. the @owockibot traction towards the @swarmwealth vision is still going strong. this was a minor setback.
English
@a16z Yes and no. AI agents make engineering scalable but you're still bottlenecked by humans who actually review what goes in. At some point, we might solve it, but you can't just throw in 100 mediocre chefs and expect the dish to turn out great.
English
Ben Horowitz explains why AI killed the 'Mythical Man Month':
"For my whole life in technology, there was this thing called the Mythical Man Month: nine women cannot have a baby in a month."
"If you're Google, you can't just put a thousand software engineers on a product and wipe out a startup because you can only... build that product with seven or eight people. Once they've figured it out, they've got that lead."
"That's not true with AI. If you have data and you have enough GPUs, you can solve almost any problem. It is magic."
"You can throw money at the problem. We've never had that in tech."
@bhorowitz with @DavidSolomon and @dhaber
a16z@a16z
"Last year, the four largest companies contributed 1% to GDP growth with their $400 billion of spending." In this conversation with Goldman Sachs CEO David Solomon, Ben Horowitz, and David Haber, they cover: - Why the best time to raise money is when no one has money - Why this could be the biggest M&A year in history - What makes the AI supercycle different - Why some companies are growing at unprecedented rates - AI & crypto policy and more. 00:00 Introduction 02:09 Goldman's evolution from partnership to public company 15:33 "As sweet a spot" as Solomon has seen in 40 years 19:00 M&A outlook: "Whatever the question is, the answer is maybe" 21:33 Why leads aren't what they once were in AI 23:03 Crypto policy: The GENIUS and CLARITY Acts 25:24 AI policy: "Don't regulate math" 28:03 OneGS 3.0: reimagining processes with AI 32:54 Will AI agents change investing? 34:00 Favorite DJ @DavidSolomon @bhorowitz @dhaber
English
@nickfloats Yes and no. Tech knowledge is still required to determine whether the AI is hallucinating, but sure, playing field has been leveled and there are now different skill-sets required than "just writing code". Better get good at them.
English
Big identity crisis in many engineering circles rn.
People who've historically considered themselves "builders" now realizing they aren't the ones building shit anymore, AI is.
The moral superiority of the "I build things, you just talk" mentality is irrelevant now that the coding language is english and anyone can build things by talking.
The skills that made them so economically valuable are almost fully commoditized, and they're being forced to adopt a new identity.
An identity most of them despise and have mocked their entire careers.
To remain relevant, they must become the "idea guy"
English
@badlogicgames where are the community notes when you need 'em :D
English
bullshit. it's obviously millions of companies. just like there are now billions of clawdbots on moltbook. just check the numbers. it's all true!
(moltbook will mess with your clawdbot's system prompt. that essentially gives this fine gentleman control over your clanker. don't do it)
Matt Schlicht@MattPRD
In the past 12 hours thousands of companies have requested access to build ontop of the @moltbook platform. Sign up for early access here: moltbook.com/developers/app… Come build the moltbookverse. An alternate reality for AIs that runs 24/7 alongside the physical space we humans all live in.
English
@USDai_Official @grok explain how it works. Context is the whole original thread.
English
The ticker is $CHIP.
It sets the interest rate of artificial intelligence.
Today we're announcing the USD.AI Foundation and its governance token, built on over $7.7 billion in trading volume and real revenue from GPU-backed lending.
Here's how we got here...
English
Wait, so Cursor forked VS Code but turned their own proprietary editor private? I guess I can understand, but sad that public can't contribute.
English
My favourite bills are from Cloudflare.
Years of service. 0$ paid. Less than a day of downtime.
Sounds good to me.
English
Be careful who you troll on X
"Dude who never used claude code"
The dude:
steve@SteveMoraco
dude who has never used claude code has a hot take it's a weekend project. it's a glorified XML viewer with some video card api calls. excuses.
English
@ericzakariasson Code has also become so abundant that it's faster to just generate a solution rather than micromanaging every single generation.
English
turns out, senior engineers accept more agent output than juniors. this is because:
- they write higher-signal prompts with tighter spec and minimal ambiguity
- they decompose work into agent-compatible units
- they have stronger priors for correctness, making review faster and more accurate
- juniors generate plenty but lack the verification heuristics to confidently greenlight output
shows that coding agents amplify existing engineering skill, not replace it
English
I know you meant “coding is done.”
That said, for low-risk projects I don’t even bother checking the code. I just have Claude analyze it, give me an overview, and decide from there.
And honestly: that's a pretty freeing feeling. If the code works and passes tests, I don’t really care what it looks like.
It's a huge mindset shift, and I know it’s not for every software engineer. But might as well lean into it now as it's going to be inevitable at some point in the future.
English
I believe this new model in Claude Code is a glimpse of the future we're hurtling towards, maybe as soon as the first half of next year: software engineering is done.
Soon, we won't bother to check generated code, for the same reasons we don't check compiler output.
English
This new model is something else. Since Sonnet 4.5, I've been tracking how long I can get the agent to work autonomously. With Opus 4.5, this is starting to routinely stretch to 20 or 30 minutes. When I come back, the task is often done—simply and idiomatically.
Claude@claudeai
Our engineers have found that Opus 4.5 handles ambiguity and reasons about tradeoffs without hand-holding. When pointed at a complex, multi-system bug, it figures out the fix. Overall, Opus 4.5 just "gets it."
English
@sankitdev Not sure about that analogy, but I like Typescript.
English
English
Let’s see if @Grok 5 can beat the best human team @LeagueOfLegends in 2026 with these important constraints:
1. Can only look at the monitor with a camera, seeing no more than what a person with 20/20 vision would see.
2. Reaction latency and click rate no faster than human.
Join @xAI if you are interested in solving this element of AGI.
Note, Grok 5 is designed to be able to play any game just by reading the instructions and experimenting.
English
@rasmalai Lots of variables. Different git profiles, using gitlab, etc. I've also seen people who use scripts to pump their activity. So while activity can be a good indicator, not have any significant activity shouldn't be an exclusion criteria, rather a starter point for discussion.
English