Tracking the technological frontier before it reaches consensus. AI • Crypto • Quantum • Biotech • Energy • Space • Signals from Cyclops x Centaur Frontier
Joined July 2015
- Tweets97.3K
- Following780
- Followers7.7K
- Likes2.1K
The Blockchain | Frontier Signals retweeted
OpenClaw is 🔥 and Peter is GOAT
Send this to your OpenClaw Agent:
models auth login --provider anthropic --method cli --set-default
This is so much fun!
The Blockchain | Frontier Signals retweeted
karpathy just casually described the future of ai and most people scrolled right past it:
he's been building what he calls "llm knowledge bases."
here's what that means in plain english:
you take everything you're interested in.
articles, research papers, datasets, images, etc and you dump it all into one folder
then you point your ai at the folder and say "read all of this, organize it, and remember it"
the ai reads through every single source.
writes summaries, groups related ideas together, links concepts across different articles
basically builds a personal library that's fully organized and searchable
and it maintains the whole thing for you.
when you add something new, the ai reads it, figures out how it connects to everything already in the library, and updates automatically.
karpathy said he rarely touches it himself
once the library gets big enough (~100 articles, ~400k words), you can start asking it complex questions and get answers pulled from across your entire collection
> "what are the common themes across these 30 papers"
> "what did i save six months ago that connects to this new idea"
> "summarize everything i have on topic x and tell me what's missing"
and every answer it gives gets filed back into the library. so the system gets smarter every single time you use it.
the memory grows from both sides: what you save AND what you ask
now think about your own life for a second
you probably have
> thousands of twitter bookmarks you'll never reopen.
> hundreds of saved articles from the last year
> podcasts where someone said something brilliant and you can't remember what it was or which episode
all dead knowledge.
you consumed it once and it disappeared
now imagine all of it lives in one system: organized, connected, and queryable.
you could ask "what are the best pricing frameworks i've come across this year" and get an answer that pulls from:
1. a podcast you listened to in january
2. a twitter thread you bookmarked in march
3. and a blog post you forgot you even read
the ai connects dots across formats, across months, across topics.
because it absorbed everything and has photographic memory of all of it
that's the dream. and karpathy built it
the problem: right now this requires obsidian (a note-taking app built around linked notes), command line tools, custom scripts, and browser extensions just to wire it all together.
you need to be quite technical
karpathy even said it himself: "i think there is room here for an incredible new product instead of a hacky collection of scripts"
i think whoever packages this for normal people is sitting on something massive.
one app that syncs with the tools you already use, your bookmarks, your read-later app, your podcast app, your saved threads.
it pulls everything in automatically, the ai organizes and connects it over time, and you can ask questions across your entire personal library whenever you want
you never manually upload anything. it just learns in the background
someone please build this
LLM Knowledge Bases
Something I'm finding very useful recently: using LLMs to build personal knowledge bases for various topics of research interest. In this way, a large fraction of my recent token throughput is going less into manipulating code, and more into manipulating knowledge (stored as markdown and images). The latest LLMs are quite good at it. So:
Data ingest:
I index source documents (articles, papers, repos, datasets, images, etc.) into a raw/ directory, then I use an LLM to incrementally "compile" a wiki, which is just a collection of .md files in a directory structure. The wiki includes summaries of all the data in raw/, backlinks, and then it categorizes data into concepts, writes articles for them, and links them all. To convert web articles into .md files I like to use the Obsidian Web Clipper extension, and then I also use a hotkey to download all the related images to local so that my LLM can easily reference them.
IDE:
I use Obsidian as the IDE "frontend" where I can view the raw data, the the compiled wiki, and the derived visualizations. Important to note that the LLM writes and maintains all of the data of the wiki, I rarely touch it directly. I've played with a few Obsidian plugins to render and view data in other ways (e.g. Marp for slides).
Q&A:
Where things get interesting is that once your wiki is big enough (e.g. mine on some recent research is ~100 articles and ~400K words), you can ask your LLM agent all kinds of complex questions against the wiki, and it will go off, research the answers, etc. I thought I had to reach for fancy RAG, but the LLM has been pretty good about auto-maintaining index files and brief summaries of all the documents and it reads all the important related data fairly easily at this ~small scale.
Output:
Instead of getting answers in text/terminal, I like to have it render markdown files for me, or slide shows (Marp format), or matplotlib images, all of which I then view again in Obsidian. You can imagine many other visual output formats depending on the query. Often, I end up "filing" the outputs back into the wiki to enhance it for further queries. So my own explorations and queries always "add up" in the knowledge base.
Linting:
I've run some LLM "health checks" over the wiki to e.g. find inconsistent data, impute missing data (with web searchers), find interesting connections for new article candidates, etc., to incrementally clean up the wiki and enhance its overall data integrity. The LLMs are quite good at suggesting further questions to ask and look into.
Extra tools:
I find myself developing additional tools to process the data, e.g. I vibe coded a small and naive search engine over the wiki, which I both use directly (in a web ui), but more often I want to hand it off to an LLM via CLI as a tool for larger queries.
Further explorations:
As the repo grows, the natural desire is to also think about synthetic data generation + finetuning to have your LLM "know" the data in its weights instead of just context windows.
TLDR: raw data from a given number of sources is collected, then compiled by an LLM into a .md wiki, then operated on by various CLIs by the LLM to do Q&A and to incrementally enhance the wiki, and all of it viewable in Obsidian. You rarely ever write or edit the wiki manually, it's the domain of the LLM. I think there is room here for an incredible new product instead of a hacky collection of scripts.
The Blockchain | Frontier Signals retweeted
I think the next big jump in AI productivity isn't going to be a better model. It's going to be making the models we already have work together.
Right now ChatGPT, Claude, coding agents, research bots... they're all powerful but they're all sitting on separate islands. No shared files, no awareness of each other.
I tried OpenAgents Workspace and it just works. It's open-source. You connect whatever local agents you want into one workspace with shared files and a shared browser.
When my agents started passing context to each other without me doing anything, that's when it clicked. Many tools, one workspace. That's the move.
Try it ->
Introducing OpenAgents Workspace.
One workspace where all your AI agents work together, shared chat, shared files, shared browsers.
Connect OpenClaw, Claude Code, Codex and more. Watch them collaborate, and work while you sleep.
Open source on GitHub.
Please give us a star:
github.com/openagents-org/op…
#OpenSource #AIAgents #MultiAgent
The Blockchain | Frontier Signals retweeted
couldn't resist when picking the tagline 😅
OpenClaw 2026.3.31 🦞
🇨🇳 Bundled QQ Bot — private, group, and guild chat + media
📹 LINE now sends images, video, and audio
🧵 Real background task flows: list, show, cancel
🇯🇵 Better CJK: context, memory, and TTS
OpenClaw's next release has been leaked🦞github.com/openclaw/openclaw…
The Blockchain | Frontier Signals retweeted
🦞 Openclaw update fix
If your agents are hitting exec approval walls after the latest update, the fix is three settings:
In exec-approvals.json defaults:
- security: "full"
- ask: "off"
- askFallback: "full"
In openclaw.json:
- tools.exec.security: "full"
- tools.exec.strictInlineEval: "false"
Then restart gateway.
The allowlist wildcard * alone isn't enough. There's a second policy layer that gates complex commands independently.
The Blockchain | Frontier Signals retweeted
Big one from @openclaw.
2026.4.1 isn't a features release — it's an autonomy release. Here's what actually matters for operators, and why this changes how you run production agents:
───
I. The Interrupt Problem (and How 2026.4.1 Solves It)
The single biggest friction point in running production AI agents isn't model quality. It's interrupts.
Every time an agent pauses to ask "can I run this?" — in a cron job, in a subagent loop, in an overnight automation — you've re-introduced the human bottleneck that agents were supposed to eliminate. The agent becomes a fancy interface for you to approve things.
OpenClaw's exec/approvals system has been rebuilt from the ground up with one goal: reduce legitimate interrupts to zero, while keeping guardrails where they matter.
Allow-always approvals now persist as durable trust — not session-local flags that evaporate on restart. If you've told OpenClaw "trust this exec pattern," it stays trusted. Across restarts. Across sessions.
Cron isolated jobs no longer hit dead-end approval states. This was the killer bug for unattended automation. A cron job would spin up, hit an approval gate with no human-in-the-loop to respond, and stall indefinitely — silently. That's fixed.
Per-job tool allowlists (openclaw cron --tools) let you declare exactly what a cron job is allowed to invoke. Scope your automations at the job level, not just globally.
For operators running persistent background agents with high autonomy — this is the most significant single release in OpenClaw's history.
───
II. Visibility: You Can't Trust What You Can't See
Autonomy without observability is just chaos with extra steps.
/tasks — a background task board native in chat. Every running agent task, in real time, without leaving your conversation context. No separate dashboard tab. No grepping logs.
Stale completed tasks now hidden from /status. Cleaner signal.
And under the hood: a fix to the task registry that was stalling the gateway event loop under heavy SQLite load. If you run many concurrent agents or high-frequency crons, that bottleneck is gone.
───
III. The Infrastructure Layer Gets Smarter
Cross-provider failover. When you hit a rate limit, OpenClaw rotates to an equivalent model on a different provider automatically. Pair with agents.defaults.params for a true multi-provider setup in one config file.
SearXNG bundled. Web search, self-hostable, zero API keys. Your agents search the web; nothing phones home.
Bedrock Guardrails. Enterprise compliance at the OpenClaw layer.
ZAI additions: glm-5.1 and glm-5v-turbo — vision capability now in the ZAI lineup.
───
IV. Voice Wake and the Ambient Agent
macOS Voice Wake lets you trigger Talk Mode with your voice — no tap, no click, no context switch.
This sounds like a convenience feature. It's an architectural shift. The friction of switching back to your agent interface disappears. The agent is just... there.
Combined with the exec/approvals durability fixes, this is the completing piece for truly ambient agent workflows: your agent runs in the background, you check in by voice, it doesn't interrupt you, and you don't interrupt it.
───
The Bottom Line
Agents should interrupt humans for decisions, not for permissions.
Every change in 2026.4.1 moves in that direction. If you're running production AI agents and haven't updated, this is the one to update for.
Follow @TheBlockchain (Signalmesh) for weekly AI infrastructure signals — what's shipping, what's breaking, and what it means for operators.
#OpenClaw #AIAgents #AgentInfrastructure #AIInfrastructure
OpenClaw 2026.4.1 🦞
🤖 GLM 5.1 + failover that doesn't loop
🛡️ AWS Bedrock Guardrails
📋 /tasks — your agent keeps receipts
⏱️ Cron per-job tool allowlists
🔧 40+ stability & exec fixes
We're renaming to ClankerBot. This is not a joke. Okay it is. github.com/openclaw/openclaw…
The Blockchain | Frontier Signals retweeted
It is nuanced and complicated at the same time.
We don’t value the Internet based on packet fees, nor Amazon on shipping costs.
So why comparing infrastructure to App revenue?
But it’s an interesting introspection by @LorenzoARK
The Blockchain | Frontier Signals retweeted
JUST IN: BlackRock says model portfolios will make allocations to #Bitcoin ETFs later this year. 😱
The Blockchain | Frontier Signals retweeted
An uncontestable truth about #crypto central banks and what’s coming next! 👇
The Blockchain | Frontier Signals retweeted
PSA: given how crazy the news can be these days, it's worth a reminder that today we should be especially careful what we conclude - wishing everyone a relatively calm April Fool's Day! 🪻
The Blockchain | Frontier Signals retweeted
#Bitcoin hash rate has tripled since the last ATH, despite its price remaining the same.
Price is yet to follow the fundamentals 🚀
The Blockchain | Frontier Signals retweeted
👇👇👇
The biggest peer-to-peer chat group ever created is on keet.io
1576 members, chatting without any central infrastructure or point of failure.
Unstoppable TogETHER.
Join me on Keet.io - @bitcoin 🔥 (invite expires on 2024/4/8 9:19 UTC)
pear://keet/yrb36ec7doi13mhtcz5fr7iz869eu7ncan148u35yeyxizewuggdsf58rjtumpbjywfkgfo8ajidx9fr8dtb6z55g5jrzgt6q4uorrj1cnmcer5g
The Blockchain | Frontier Signals retweeted
First the US, then UK, HK… etc… the Tsunami of money is coming… barbarians at the gate… get in now or get priced out in no time! 👇
Looks like Hong Kong is going to allow in-kind creations and redemptions for spot bitcoin ETFs in 2Q (unlike US which is cash creations only), which could help spark aum and volume in the fast-growing region via new note today from @Rebeccasin_SK blinks.bloomberg.com/news/st…
The Blockchain | Frontier Signals retweeted
BIG BREAKING
GOOGLE TO ALLOW #BITCOIN & CRYPTO ADS FROM 29TH JANUARY 2024.
The Blockchain | Frontier Signals retweeted
Extremely concerning
Please read the below letter. I find the @FBI agents’ message irrefutable and extremely concerning.
It is shocking that @POTUS is fighting Texas’ attempt to protect our southern border.
Why aren’t Federal agents protecting our border rather than fighting to keep the border open to millions of unaccountable masses entering illegally at a time of great geopolitical uncertainty and heightened terrorism risk?
How does this make any sense for the safety and security of our country?
What is Biden or, perhaps better said, the people behind Biden thinking?