Welcome to the AI Untapped Workshop. This is the room behind the newsletter and the YouTube channel. The whole point: cut through the AI noise, faster, together. Here's what lives here: → One new tool + prompt every Sunday. Posted in the feed. Copy, paste, run it. → The Field Notes: five short PDFs covering local AI setup, prompt patterns, the skip list, the maker stack, and the Sunday routine. Find them in the Classroom tab (top nav) or scroll the feed today, where each one is posted separately so you can ask questions in the comments. → An open thread where you drop the AI tools you're testing and what's actually working. → Behind-the-scenes context I don't put on YouTube. What I'm asking from you: → Post what you're testing. Wins and failures both. The room learns faster when we're honest. → No pitching. No "I built an agency, DM me." This is a workshop, not a marketplace. → If a prompt works, drop the result in the comments. If it breaks, drop the error. I'm here daily. Drop a hello below and tell me one thing you're trying to figure out with AI right now. I read every reply. — Dusty

This one slipped under the radar this week and I wanted to put it here because barely anyone is talking about it. On June 2 the White House signed a new executive order on AI called Advanced AI Innovation and Security. The short version is that the US wants to stay at the front of AI, and to do that they are setting up a framework around the most powerful frontier models. A few things stood out to me. It sets up an AI cybersecurity clearinghouse and a voluntary framework for the government and the AI labs to work together on developing and deploying frontier models. It also points toward the government getting early access to evaluate the most capable models for security reasons before they go out. And it is written as voluntary, with no mandatory licensing or pre approval required, at least for now. The part I keep going back and forth on is that word voluntary. On paper it is just a foundation and the labs are not forced to do anything. But once national security becomes the framing, you have to wonder how voluntary it really stays, and what that means for whether these companies keep anything open source down the line. I broke the whole order down in a new video. What it actually says, the national security angle, and where I think this could go with the big labs. Link is here: https://youtu.be/nlmrJKmPgxM Curious what you all think. A real guardrail, or the first step toward the government having a say in every frontier model? Drop a comment, I read all of them.

Anthropic dropped Opus 4.8 yesterday. The benchmark gains are real but not the story. The story is one paragraph in the release notes that most coverage skipped over. Claude Code now has a primitive called dynamic workflows. Here is what it does. One Opus session reads your task, decides how to decompose it, spawns hundreds of subagents in parallel, audits each one's output before accepting it, and merges everything into one delivered artifact. For code, that artifact is a PR ready for human review. The example Anthropic flagged is a codebase migration spanning hundreds of thousands of lines, taken from kickoff to merge inside a single Opus session. That is the test prior subagent patterns could not pass without you hand-stitching the orchestration code yourself. Why this matters more than the benchmark bumps: The model is now the orchestrator. You stop being the project manager for your own subagents. The session still runs inside Claude Code's permission model with hooks and gates intact, so safety boundaries hold. What disappears is the spec-each-subagent step that ate the most setup time. A few other things worth holding in your head: Pricing is unchanged from 4.7. Same dollar per million tokens. Fast mode (the 2.5x speed variant) is now roughly three times cheaper than equivalent settings on prior generations. Agentic coding score moved from 64.3 to 69.2 percent. Honesty improved measurably. Anthropic says 4.8 is about 4x less likely than 4.7 to let a code flaw pass unflagged. That number is the lever that makes the parallel subagent pattern safe to actually use. Two Opus releases in 41 days at fixed pricing. If you've been waiting for the moment to move serious work onto Opus, this is the cleanest one in months. I put together a 3-page briefing PDF if you want the full numbers and the cadence read. It's in the resources folder.

This is the single most underused prompt I know. After Claude gives you anything (a strategy, a draft, an analysis), don't accept it. Send this: "List the 3 biggest weaknesses in what you just wrote. Then rewrite addressing them." That's it. Two lines. Output gets sharper every single time. I use this on every Pine Script strategy Claude writes for me in TradingView, and the v2 is almost always tradeable where the v1 was curve-fit garbage. Reply with the most surprising thing Claude found wrong with its own work. I'll feature the best one in this week's post. Dusty

Where this room is heading: NOW (this month): → Weekly tool + prompt drops every Sunday → Field Notes archive available to every member → Open Q&A — drop any AI question in the feed, I'll answer → The newsletter ships every Sunday (aiuntapped.ai, 5-min read) NEXT (the coming weeks): → First live workshop call. Bring your stuck prompts and we debug them together. Free for everyone who's in the room early. → Skool Games entry — when this room hits 100 active members, we compete. → A paid tier ("AI Untapped Pro") for people who want the full stack: pre-built workflows, working Claude projects, weekly office hours, and the behind-the-scenes archive. Not launching yet. Free tier stays free, forever! The free room is the front door. It's not a stepping stone you have to leave. Stay here as long as it's useful. One ask: reply below with the AI problem you wish someone would solve for you. I read every reply, and the most common answers shape what gets dropped next. — Dusty