Be Weird - by Joshua Tiernan - Tiny Empires
Tiny Empires
SubscribeSign in
Be Weird<br>Doing the opposite is now a strategy
Joshua Tiernan<br>May 15, 2026
Share
I get about ten cold emails a day now that are obviously written by AI. They’re long. They open with a congratulatory line about something I posted three weeks ago. They drop in a fact about my business that took the model two seconds to find and is supposed to look like research. They end with a soft call-to-action and a P.S.<br>A year ago, this kind of email worked. The personalization felt like effort. The length felt like thoughtfulness. Now the same format is being sent to thousands of people by hundreds of senders, and the result is that all of them get archived without being read.<br>Meanwhile, the one cold message I replied to last month was 8 words long. It was obviously not AI.<br>AI models by default are broad in their execution and end up being more generic.<br>Doing the opposite of what the model would do is now a competitive advantage.<br>Being weird and unexpected, is what makes something sound human. Potato.
This issue is sponsored by
You shipped the app. Apple owes you money. Why wait 45 days to see it?<br>Most App Store developers don’t realize their earned revenue sits locked for 30-45 days before Apple pays out. That’s cash you can’t use for ads, features, or tools right when momentum matters most.<br>AppCap advances up to 85% of your earned App Store revenue within 48 hours after approval. No equity given up, no long-term debt, and no commitment.<br>Rates start from 0.34% for a 5 day advance.<br>Unlock your App Store revenue →
What AI converges on
AI models are trained on roughly the same data, optimized against roughly the same objectives, and tuned to produce output acceptable to as many people as possible. The result is competence with a particular shape.<br>The output tends to be long and polite to the point of obsequiousness, heavily personalized in a way that signals research without containing any actual insight, structured into three or four neat paragraphs, emotionally warm, and slightly bland.<br>Once you can see the shape, you can see it everywhere. Cold emails, LinkedIn DMs, customer support replies, landing page copy, the follow-ups after a meeting. They all share the same skeleton, regardless of who sent them.<br>Our brains adapt to patterns quickly. The first time you got a personalized AI cold email, you might have read it. The hundredth time, you recognize the pattern in the first sentence and delete it. The pattern itself becomes a deletion trigger.
Cold outreach: do the opposite
If the AI default is a 200-word email with three personalized hooks and a soft CTA, the opposite is a one-line message with no hooks and a direct question.<br>“Want to talk?” “Worth a 15-minute call?” “Are you the right person for this?”<br>I’m not saying this as a blanket rule. In a few months time, The AI may have worked this out and started doing short emails.<br>Then it’s time to go back to doing long emails. There’s a certain opportunity in being reactive to the AI trends and going in the opposite direction.<br>A useful test before sending any outreach. If a competent AI would produce something similar from your prompt, send a different message. The exception is when the AI version is genuinely what you want to send, in which case at least cut it to half the length.
LinkedIn DMs
LinkedIn is currently the worst-affected channel. Almost every connection request and DM I receive opens with a sentence about my work that was clearly generated, followed by a pitch.<br>The opposite move on LinkedIn is to send a message that obviously couldn’t have been generated. Reference something specific that’s not on my public profile. Ask a question that requires you to have actually read the thing you’re claiming to have read. Or skip the warm-up entirely and just ask what you want.<br>A message that says “Quick question, do you handle X yourself or have someone for that?” gets more replies than a message that opens “I’ve been following your incredible journey and was inspired by your recent post about...”.
Customer support
Customer support is somewhere most solo founders are now competing against AI-generated empathy. The “I completely understand your frustration and I’m so sorry to hear you’re experiencing this issue” reply has become the standard, and customers have started recognizing it as nothing.<br>The opposite move is to drop the performed empathy and just solve the problem. A reply that says “Yeah this is broken, here’s the fix, sorry about that” lands better than three paragraphs of validation. Customers can tell when they’re being managed rather than helped, and they prefer the second.<br>This works particularly well for solo founders because you can credibly sign emails with your own name and say “I built this, here’s what’s happening.” Larger competitors can’t do that. The personal signature on a brief, useful reply is now a moat.
Landing pages
Landing page copy is converging...