Maybe important to repost this now that everyone is getting hacked If you have a VPS, I recommend you to get if off the public internet and into a Tailscale network and Cloudflare Tunnel ASAP It doesn't matter you have SSH key login, this is about exploits and vulnerabilities AI is finding There's 2 ways into a VPS: - via your site/app (like the PHP exploit of this week with a file upload hack) - via your server (like hacking into SSH or some open port) The site/app side you put on a Cloudflare Tunnel and then with the default Free Plan it has WAF security enabled by default and they jump on new security exploits fast, it means that the only way to reach your website is via the Cloudflare Tunnel, your server does not respond to port 80 or 443 anymore The server side you put on a Tailscale network which is like your own private network that nobody can access but your own devices (like your laptop, your phone and your VPS) kinda like a VPN Then you switch off ANY regular access to your VPS on the firewall, in my case I use Hetzner and you create a new firewall then empty it and that blocks everything (test it) How do Tailscale and Cloudflare Tunnel get in then? They don't. They go outbound. They make a tunnel to outside which sends and receives Now you've reduced your attack vector to Tailscale and your Cloudflare Tunnel, much safer, not perfect btw, but nothing is

We’re opening a research preview of Claude Duet, a new app where Claude works as a proactive agent. It comes to you with what needs your attention, completes tasks you ask of it, keeps working when you step away, and checks with you before it acts on your behalf.

On my way to the office this morning, I passed a group of young people protesting, a Palestinian flag draped over their shoulders. And I asked myself a question I can't shake: why does this generation always fall into the same trap? No matter the history. No matter the context. Even the ones from wealthy families, who went to the best schools, who were dealt every good card. Always the same trap. I don't think it's stupidity. I think it's a mechanism. And a mechanism can be taken apart, piece by piece. A. The observation: everything the state touches becomes unaffordable. Look at economist Mark Perry's famous chart, the one people call "the chart of the century." It tracks US prices since 2000. At the top, what explodes: hospital services, college tuition, textbooks, childcare, housing. Increases of +100%, +150%, up to +250%, far above general inflation. At the bottom, what collapses: TVs, software, phones, toys, clothing. A TV today costs a fraction of what it did in 2000, for an infinitely better product. Look at what separates the two groups. It isn't technology. It isn't complexity. It's one variable: how much the state is involved. At the top, sectors that are subsidized, regulated, shielded from competition, where the state funds demand and blocks supply. At the bottom, sectors where companies fight every single morning to serve you better and cheaper. B. The mechanism: why it's systematic. When the state subsidizes demand without freeing supply, prices rise. Every time. Guaranteed student loans, and universities raise tuition. Housing aid plus blocked building permits, and rents skyrocket. Administered healthcare, and bureaucracy eats the budget before it ever reaches the patient. In France it's even more vicious, because the price doesn't disappear: it hides. You never see the hospital bill, you pay it through payroll contributions. You never see the cost of school, you pay it through taxes. Public spending at 57% of GDP, one of the heaviest tax burdens in the world, more than €3.4 trillion in debt. And the result? Overflowing ERs, falling PISA scores, young people who can't afford to live in the city where they work. The market didn't fail in these sectors. It was never allowed in. C. Suffering demands a culprit. A 22-year-old lives all of this in their bones. They can't find a place to live. Their degree either buries them in debt or opens no doors. The future feels blocked. That suffering is real, and I don't deny it for a second. But human beings can't stand suffering without a culprit. They need a face. A scapegoat. And this is where everything is decided: who taught them how to read the world? D. Whoever tells the story picks the villain. Schools are public. Universities are public. Culture runs on grants. Film goes through funding committees. Public broadcasting shapes the narratives. The people handing down the interpretive framework are, overwhelmingly, paid by the very system that creates the problem. Do you really think that system is going to name itself as the culprit? No. It names the capitalist. The boss. The landlord. "The rich." The multinational. The commercial West. Same role, same casting, every time. E. One lens, a thousand causes. And this brings me back to that flag. A generation taught to read every situation through a single lens, oppressor on one side, rich, Western, capitalist, oppressed on the other, will apply that lens everywhere. To housing. To climate. To conflicts on the other side of the world. The cause changes, the lens stays. That's why the trap is always the same: it's not the cause that draws them in, it's the way of thinking that was installed in them. F. The perfect inversion. Here's the irony. Capitalism is the force that drove down the price of almost everything it touched. The state is the force that blew up the price of almost everything it captured. And young people blame the first for the crimes of the second. It's like blaming the firefighter for the fire because you saw him at the scene. G. The ballot box confirms it. In the US, a few weeks out from the midterms, a large share of urban young voters is getting ready to vote Democrat for more state, more spending, more regulation. New York elected a self-described socialist mayor, carried by young voters, on promises of rent freezes and city-run grocery stores. Rent freezes, the policy that has made housing scarcer everywhere it's been tried. The cure that worsens the disease. In France, it's the same play in two acts. Part of the youth votes for the radical left, which promises more state. Another part votes for the center, which promises to "manage" the state well, and has dug the debt deeper than ever. Two different ballots, one shared assumption: the problem is the market, the solution is the state. H. The way out. These young people aren't stupid. They're right to be angry. They're just aiming at the wrong target. The day they understand that their rent, their tuition and their hospital are expensive precisely because the state runs them, their anger will change direction. They'll stop protesting against those who create. They'll start protesting against those who capture. That day, the greatest intellectual scam of our time will be over.

If you learn AI Agent Engineering, you will never be unemployed again. I’d do this. Stage 1: Agent Architecture & State Machines - Learn: Directed Acyclic Graphs (DAGs), cyclic workflows, LangGraph, Temporal dot io, state serialization. - Practice: Build a multi-step agent using a state machine that can loop, branch and pause without using a fragile `while True` loop. - Why: Chat loops crash in production. Real-world workflows require deterministic state management and crash recovery. Stage 2: Tool Use & MCP (Model Context Protocol) - Learn: JSON-RPC, strict Pydantic schemas, MCP server/client architecture, dynamic tool discovery. - Practice: Write a raw MCP server from scratch exposing a local Postgres DB, handling auth and scoped permissions without using a wrapper SDK. - Why: Agents need standardized, secure hands to do real work. Bespoke API wrappers do not scale across an enterprise. Stage 3: Computer Use & GUI Automation - Learn: DOM parsing, Set-of-Mark prompting, coordinate mapping, screenshot diffing, Playwright/Puppeteer. - Practice: Build a vision-based agent that logs into a legacy web portal, navigates menus via screenshots and downloads a CSV. - Why: 80% of enterprise software has no API. GUI agents unlock the legacy world that text-based agents cannot touch. Stage 4: Advanced Memory & Context Engineering - Learn: 3-tier memory (working, episodic, semantic), vector eviction policies, context compression, token budgeting. - Practice: Build a memory engine that automatically compresses a 50-turn conversation into an episodic summary to prevent context overflow. - Why: Infinite context causes hallucinations and OOM errors. Memory must be actively managed, scored, and evicted. Stage 5: Multi-Agent Swarms & A2A Handoffs - Learn: Agent-to-Agent (A2A) protocols, typed state passing, supervisor vs. peer-to-peer topologies, consensus voting. - Practice: Build a 3-agent swarm (Researcher, Coder, Reviewer) that passes strictly typed JSON state, never raw text. - Why: Single agents fail on complex tasks. Swarms divide and conquer, but raw text handoffs cause context drift and hallucinations. Stage 6: Sandboxed Execution Environments - Learn: E2B, WebContainers, gVisor, Docker-in-Docker, resource limits, network egress blocking. - Practice: Build a code-interpreter agent that writes and executes Python in an isolated microVM with zero internet access. - Why: Letting an LLM run `os.system()` on your host machine or production server is a catastrophic security risk. Stage 7: Inter-Agent Security & Guardrails - Learn: Indirect prompt injection, tool poisoning, sanitization proxies, least-privilege scoping. - Practice: Build a middleware that strips malicious instructions from an email before the "Email Reader" agent passes it to the "Database" agent. - Why: Multi-agent systems create internal attack surfaces. Agent A can be tricked into hijacking Agent B's tools. Stage 8: Human-in-the-Loop (HITL) & Breakpoints - Learn: Asynchronous approvals, webhook routing, Slack/Teams integrations, durable execution pauses. - Practice: Build an agent that drafts a $10k invoice, serializes its state, and waits 3 days for a human to click "Approve" in Slack before resuming. - Why: High-stakes automation requires human accountability. Agents must know how to pause, wait and resume cleanly. Stage 9: Self-Correction & Reflection Loops - Learn: Critic agents, traceback analysis, retry with context, compiler error parsing. - Practice: Build a coding agent that reads its own stack traces, rewrites the failing code and retries up to 3 times before escalating to a human. - Why: Agents that fail on the first try are useless. Agents that debug themselves are digital employees. Stage 10: Trajectory Evals & CI Gates - Learn: Step-level grading, golden DAGs, trajectory distance metrics, LLM-as-a-judge for tool calls. - Practice: Build a CI pipeline that replays 100 agent trajectories and blocks the PR if the tool-call sequence deviates from the golden path. - Why: Grading the final text output hides the fact that the agent hallucinated an API call or skipped a safety check to get there. Stage 11: Graceful Degradation & Fallbacks - Learn: Circuit breakers, API-to-browser fallbacks, model cascading, cost-kill switches. - Practice: Build a routing layer where if the primary REST API 404s, the agent seamlessly spawns a headless browser to finish the job. - Why: The real world is messy. Agents must adapt to broken environments and degraded APIs without crashing the workflow. Stage 12: Public Agent Teardowns & Portfolios - Learn: Architecture decision records (ADRs), cost-per-task analysis, latency profiling, public benchmarking. - Practice: Publish a teardown of your multi-agent swarm, detailing the exact cost, failure rate, state-machine design and eval methodology. - Why: Top AI labs hire Agent Engineers based on their system design judgment and production reality not their ability to import LangChain. Bookmark and Repost!

Gemini 4 Argon is here. It’s available globally today for developers and enterprises in the paid tier of the Gemini API and the Gemini platform, for Ultra subscribers in Google Antigravity, and Ultra and Pro subscribers in the Gemini app and Search. Hope you all enjoy it!

The number 1 complaint I hear from people is they just don't know wtf to do with AI agents They set up Grok Bot, Muse, Hark, Dots, Hermes and then do nothing with it So I'm launching a free project: Agent Use Cases It's a website that scans the entire internet every day and finds every use case people do with AI agents Searches across basically every site on the internet to see what people are doing with AI agents that is making them money/providing them value Use cases are organized by category and which agent they're used in Gives you the use case, end result, and prompt to steal the use case Some cool use cases I've already discovered: 1. Finding unclaimed money online 2. Negotiate phone bill down autonomously 3. Teach an agent to day trade It's completely free. Nothing pay walled. You can just go on and discover cool new use cases I'd recommend browsing then giving the link of this post to your agent and having it find use cases relevant to you Sign up for the newsletter if you want 5 new use cases in your inbox every morning too Check it out: agentusecases.ai/

Spawning 110 cloud subagents inside cloudroom.dev with one prompt 110 threads spawned in 57 seconds 10 Sonnet agents, each running 10 Haiku agents Your laptop is no longer the limit