Good morning everyone, happy Friday, and aren’t you just tired of how everyone’s “writing” is the same now?

TLDR, there was an analysis of ~1M text messages written before and after ChatGPT. The after sample shows that writing has gotten measurably less varied. While the models have kept the meaning they’ve flattened the vocabulary and soul. The classic...this isn’t bad…it’s just…not good/you/real?

I've genuinely been feeling this at work the most. AI has been a huge unlock for me as a PM and builder obviously, but oh man, the amount of things just being handed off by people that’s clearly done-by-Claude is insane.

So, I've started writing less pretty docs and plainer and more basic slides by hand like a caveman. No AI allowed in Jira tickets either.

I can't keep reading decks from Claude. It discredits the thinking even when/if the thinking is all human. I'd rather the presentation of my human ideas and thoughts be sloppier, but obviously mine, than pretty and polished but the AI slop question mark over it.

I saw a 27 slide deck the other day from an expensive agency. Zero doubt made by AI and the message was the same on almost every slide just pointed at a different audience. Instant no.

It’s counterintuitive but IMO the business lesson here is backwards from what our earlier careers taught us. Polish used to be proof of effort. A gorgeous deck was in 2022. But now it's evidence against us. Is the new proof of effort the plainer thing that says the voice and thinking is yours? Authenticity. Anyways…

In the next 5 minutes over your coffee: how to steal thunder and roast your competition, why nobody cares about your launch, why people are more scared of $10K bets than $10M ones, why software is making people insane, and why your first company name should suck.

Despite the opening ramble, I still love using AI. It’s just about the where and how and meeting notes are the perfect use case for a background AI companion.

Most AI tools still make you do the work. That’s not how it’s supposed to be. You copy-paste notes, re-read transcripts and try to remember what was said in which call by whom. 

Granola Chat works differently. It searches across all your meetings, finds the most relevant ones, and gives you real answers with citations linking back to the exact note it pulled from. Ask Granola anything from anytime:

It knows when to give you a quick answer and when to go deep across your entire meeting history. It can search the web for more context, so you're not stuck in a bubble of your own notes.

Apple's new iPhone Duo does indeed look gorgeous. Pricey but stunning. My favorite thing though was how Duolingo got in on the action.

Within minutes the owl posted "it's called the iPhone WHAT???" then "oh god oh god my SEO my SEO." Then "they named a phone after me and made it bend over." And the kill shot: "iPhone Duo: $1999. Duo on your iPhone: free." Samsung hopped in calling the Duo "reheated leftovers" and tried a solidarity post, and Duolingo replied "when an ugly stranger calls you twin, No Brand." Samsung's posts alone pulled a million+ impressions, and the roast got serious coverage.

A big launch from one company can be a marketing event for anyone paying attention. It's a public attention pool ready for wit and it’s a good example that there can be many winners. Apple won, but so did Duolingo—Apple spent months and millions and Duolingo spent about nine minutes, no budget, and several years of building a personality people already knew.

That last part is the reason Duolingo’s roasting was so good. It was lightning fast to grab the opportunity and perfectly on-brand and expected because they have a brand voice built. And that voice is what allows you to speak up at any moment without layers and layers of approval to post anything online. To be in that position and catch the wave, think about your brand's reaction voice before you need it. What does it find funny, what would it never say? Then give one person permission to post inside ten minutes with zero approvals. Relevance is a speed game 👇

Tendem is a bench of 10,000+ vetted human experts inside your AI chat. Describe the task, see a price, approve it, and a real person sends it back finished with multi-layer QA. You now get:

As someone who loves making apps and things, this makes me a little sad. But Kyle Poyar isn’t wrong. Shipping got fast but marketing, sales and customers didn’t.

Everyone is posting their next new app on LinkedIn every day, but with so many new things the volume is just cranking up and it’s infinitely harder to hear what matters. Kyle found three companies that made launches the growth lever instead of adding to the noise: Notion, Rippling, and Profound.

The release and the launch are two different decisions made at two different times. That kills the one-way-door feeling of “launch day” and means nothing has to be perfect to reach customers. Also the creative bar has split in two: beautifully crafted or authentically raw. Profound's launch video was Zoom reaction clips from beta calls. "The thing that dies is the middle."

Two things to steal from Kyle’s playbook. First, the founder-written release notes every 2–3 weeks to every customer. Second, have the people who built the individual things post and talk about the thing they made—not the company. Have Mitch and Mo who built the new job wall feature be the voice of that feature.

Because a $10M bet comes with cover. It goes through a committee, three decks and a sign-off chain, so if it fails, "we" decided it. A $10k experiment has your name on it.

“People are terrified of small bets” is the argument in a guest post on Annie Duke's Substack. Two Fortune 500 CEOs said that their teams "aren't imaginative." No, it’s that the operating system punishes anyone who stands next to a bad outcome so nobody does.

Small bets aren't scary because of what they cost but because someone has to be the person standing next to the bad outcome. Annie calls the disease "resulting"—judging a decision by how it turned out. In a hierarchy it runs one way so people learn the rational lesson: never place a bet you can be blamed for. So, bets either don't get placed or they become zombie projects nobody will kill.

The only way to get rid of that fear is to make making a good bet the path of least resistance so people don’t mind making it. One tactic is to pre-commit to what the kill switch is. For instance no adoption by X, no customer signal by Y, then when it crosses the line "kill it with pride and pocket the data."

An essay by GrayBeard did the rounds this week on Hacker News. The idea is that software combines speed, money, complexity, abstraction and unlimited freedom to change your mind. Separately fine things but together they turn "otherwise normal people" into "b-tier Bond villains." Because if you strip away the branding and whistles, well, GrayBeard says most software is still just a glorified spreadsheet. A form, an endpoint, some permissions, a database.

If you surround people with levers they will eventually pull them. Slow growth, change the funnel. Confused customers, redesign. Still struggling, reorg then pivot. Eventually the company itself gets treated like software: permanently mutable, permanently one refactor from working. And complexity feels important to some degree to justify being a product because a CRUD app "sounds boring" while an event-driven platform with a service mesh "sounds like NORAD." With AI agents making everything now the only thing left between you and over-engineering is proportion (and taste).

Leaving things alone is an underrated skill. Guilty guilty guilty. I can’t stop tinkering, but sometimes we must.

Yes according to Arielle Jackson (Head of Brand at First Round) who has named dozens of companies and wants you to stop agonizing about it and wasting money. Her old startup was incorporated as "Apps & Zerts, Inc." and only became Cover once they knew what they were building. Maven ran nearly eight months with no name at all.

The idea is to pick a placeholder so outlandish it could never see daylight so you don't launch with it. Then budget at least a month for the real one once you have a better idea of what the product is. And relax. Disney was Walt’s last name—"OK name, remarkable brand."

People spend a lot of time and money on their name. Finding clever ones, buying domains, doing trademarks. We all know the first move in starting a company begins on GoDaddy.

A name matters but the name is just the ambassador for the brand, not the brand. It can feel like a one-way door, but if you defer it to later it’s not. Also, she says the good names are polarizing. If everyone thinks it's fine it's because it’s the safe boring one. Nike came off a list Phil Knight hated but it grew on him.

Positioning statement → naming brief → namestorm across 8–10 themes → shortlist scored by trademark price, domain price, sound and appearance. Her post has two nice tricks. (1) don't ask AI for name suggestions, ask it for raw material like give me 25 lesser-known engine parts. And (2) tell three people your three names and ask them two days later which they remember. The full 12-question brief for naming startups is here 👇

🛠️ A product—Go Name Yourself, a deck of cards for running a product namestorm → What does a card game for startup naming look like?

📺 A video—If you haven’t seen much about the Duo yet, Marques has got you → iPhone 18 Pro/Duo Impressions: Mogged

📬 A read—Want to know how to disagree better? → Read how (and why it matters)