Good morning guys, happy Friday, and for everyone in the US congrats on making the round of 16, and I hope you have a great 4th of July weekend! 🦅🌭🎇

In the next 5 minutes over your coffee: the new Claude Code command that finishes the whole job while you’re away, the AI “salesperson” living inside Monday’s product converting trials at 2.5x, Figma’s Dylan Field on the one edge AI can’t hand you, how three YouTubers just beat Star Wars at the box office, and the AI device SpaceX swears it never showed investors.

Fable is back. Also, I have it, but I’m definitely a foreign national here? Don’t tell anyone.

And the reason to care fits in one command: /goal. Instead of handing Claude Code a task, you hand it the finished result and walk away. It plans, builds, checks its own work, fixes what broke, and repeats—while a second model grades the output against your goal so it can't fake “done.”

It's the first model with the context stamina and wherewithal to actually run a goal to done. The whole trick is the “Done when…” line, and the agent has to actually hit it. Like: “/goal: 20 ad angles for [product], each backed by 3+ real customer quotes from reviews and Reddit.” Vague goals like “make it good” just loop forever.

Save it for work that matters, and know three things going in: it's slow (it thinks before every move—I’d still grab a smaller model for quick jobs), the safety filter is jumpy in health and security-adjacent niches and can bounce you to a less capable model mid-task (if Fable suddenly feels dumber, check which model is answering), and it eats your usage limit fast—one big job can swallow a scary chunk of your weekly cap.

I built a free App Health Score—one of my 4 free tools. Drop in your numbers (conversion, trial-to-paid, churn, LTV, retention) and it grades your app 0–100, then shows exactly which metric is dragging you down, benchmarked against real data.

Built in partnership with my friends at RevenueCat, who have by far the largest dataset of app and subscription data out there.

I also built the competition heat calculator, a revenue/unit-economics calculator, and an ask-anything tool. Poke around the whole set.

The freshest GTM motion seems to be dropping an AI agent into the funnel at the exact moment of intent. Monday.com (over $1B ARR) runs three: Amanda answers 100% of English inbound and cut response time from 24 hours to under 2 minutes; Oscar does account research that used to take reps 1–2 weeks in about 5 minutes; and Jax—the one to steal—pops up during the free trial as a live “sales engineer” that can actually configure the product for you.

Jax converts trial users at 2.5x the control group, with 3,000+ calls a month and half coming back for a second. The tell that it's really working: users miss Jax when the trial ends. That's not a sentence you hear about sales automation. In-product, at the moment of intent, beats another outbound sequence every time.

Find your activation gap—the moment a trial user gets stuck or goes quiet—and put help right there, in the product, live. You don't need Monday's budget: even a scrappy in-app assistant that spots intent and walks someone to first value is the play.

A flow breaks, so someone clicks through, screenshots it, drops it in a ticket, and tries to describe what “feels off.” By the time an engineer picks it up, the context is scattered across comments and images—and a small UX fix turns into a slow, error-prone handoff.

Rezonant Alter closes that gap. PMs and designers capture the broken UX right in the product—what happened, what was expected, where it broke—and engineers get clear, ship-ready tickets with exact code references. Smaller bugs stay small.

In a new Stratechery interview, Figma CEO Dylan Field dropped a noteworthy line on building with AI: models are trained on the distribution, so the most interesting, differentiated work is out of distribution by definition. Put plainly—when everyone prompts the same models, the average gets commoditized and taste becomes the moat.

When execution is nearly free (and it essentially is despite costs going up), the edge moves to judgment: what to make and why, not how fast you can generate it. In short “is the AI using you, or are you using it?” Plenty of teams get tunnel/trust vision and just let the model pick the direction, and mistake motion for progress. Going fast the wrong way is still the wrong way.

Use AI to execute. Don’t outsource too much thinking and decision making. Before you just hit done on what the model handed you, really critique it like an annoying boss would…is it just the obvious middle-of-the-distribution answer? And push it and yourself—where you can deliberately go out of distribution: a sharper opinion, a weirder format, a real point of view. The taste-driven, hard-to-verify parts of your product are exactly the parts that stay yours.

Something interesting is happening in Hollywood. Three films that came from the YouTube world just landed in the domestic box-office top five in the same weekend—and each kept a different amount of power. The horror film, Backrooms took the acquisition route with A24: a $10M budget, $118M worldwide. Obsession (so good btw) signed a distribution deal and turned $750K into $171M. And Glitch's Amazing Digital Circus finale didn't sell to anyone—it rented the theaters and kept everything.

What’s unique/new is the mechanics on the supply side: demand now originates on platforms studios don't control where the audience assembles itself years before a greenlight. As one studio chief put it, by the time these creators make the movie, they've had “a billion test screenings.” Build the audience first and you get to choose how much of the upside you hand over.

This is why MrBeast got so much money and freedom to do Beast Games on Amazon.

Same lesson whether you ship films or software—own the audience relationship before you need it, and gatekeepers become optional. The creators with the most leverage spent years building direct fandom, then rented distribution instead of selling their business. Your direct list and your community is the the asset that lets you set the terms later.

Don't have that direct line yet? beehiiv is where I'd start building it today—it's the exact stack I run The Diff on. Send me a link to your newsletter, I’ll read it. → Start building my audience on beehiiv

Right before the IPO, SpaceX reportedly showed investors a prototype AI device—slimmer than an iPhone, its own operating system, a Qualcomm chip, and xAI's Grok baked in. The vision: one vertically integrated stack (chip → model → satellite → device) that routes around Apple and Google entirely. Musk's reply to the report? “Utterly false.” (He would say that though.)

Everyone's chasing the post-phone AI form factor—OpenAI with Jony Ive, and now maybe Musk. It's the move to own the whole stack so no platform sits between you and the user. But the AI-hardware graveyard is real—Humane's Pin ($700 + $24/mo) got switched off and sold to HP for parts. That Rabbit product thing. Hardware is hard.

You're (probably) not building hardware, but regardless,when a platform taxes you or gates your users, the longest term game is owning more of the stack. Just respect the graveyard—new form factors despite great vision usually die on the execution part.

› For couples traveling this summer — don’t let the days blur together, and don’t worry about having to journal everything down. Just use the photos you’re already taking, and capture one little moment in under 60s everyday. (Download the app, and get a movie at the end of my trip).

› beehiiv’s Summer Release is July 16 — beehiiv (the best newsletter operating system) is doing their big summer product drop on July 16. You can RSVP to catch what they’re announcing live—come see what’s coming, it’s huge (RSVP for the Summer Release).

› Nobody likes the guesswork on how you’re doing compared to others — and access to data that answers that for your product can be a chore to sift through, which is why I did it and packaged it for you in our free set of tools— the app revenue/economics calculator, an app health score meter, and the competition heat calculator from up top (See how your app compares).

› My go-to stack with the vetted tools I trust everyday — the products I genuinely lean on to build things at work, my apps, and run my newsletter business (Explore my personal app store of recs).