Launch Day Lies—Day Two Tells the Truth

Was this newsletter forwarded to you? Sign up to get it in your inbox.

I spent six months building sophisticated AI products that users tried once and forgot. Then a throwaway weekend build became the one people couldn’t stop talking about—or talking to.

First there was TLDR, an AI podcast generator: 10 weeks of work, great initial response, zero returning users. Then there was Kairos, an AI reading app: 1,000 downloads on day one, ghost town by day three. Next was Unwrite, a Grammarly alternative: 200 users, but only 50 daily interactions total.

The idea that became Monologue was different. It was barely more than a hack: Press a key, start talking, and it transcribes your voice into text. But by day two, two users were using it 200 times per day.

This is a paradox entrepreneurs know too well: We pour months into polished products with elegant architectures and thoughtful user journeys, only to watch them die on day two. Meanwhile, the hacky prototype you're embarrassed to show becomes the thing people message you in all caps about when it breaks for five minutes. After building three failures at Every—products that looked promising on launch day and faltered shortly after—I learned that launch day lies and day two tells the truth.

It didn’t matter how sophisticated or robustly featured the product was. Instead, I was solving a problem so urgent that people would use a broken prototype 100 times before lunch rather than wait for the polished version.

Building Monologue taught me something I wish I'd known six months and three dead products ago: We're all measuring the wrong things on the wrong day.

The graveyard of good ideas

When I joined Every's entrepreneur in residence (EIR) program, I came in with a plan. Find the big idea. Go all-in. Ship something meaningful. I had a list of concepts ready, and one stood out: Turn meeting notes into podcasts you could listen to while commuting. At the time, Google's NotebookLM had gone viral for turning documents into surprisingly engaging audio, so we thought we should apply the same idea in a company setting.

We called it TLDR.

I spent 10 weeks building it. I dove deep into podcast structure, studied how Gimlet Media crafted hooks, and learned about intros and outros. I was terrible at writing, so I obsessed over prompts that could generate engaging audio. By December, I had a polished product.

Launch day felt promising. Every CEO Dan Shipper wrote about it, we got interest from prospects, and people said it was "cool" and "innovative." But internally, our team didn't touch it. Externally, it didn't fit cleanly in Every's bundle model, and there were zero returning users. The harsh truth emerged: No one wanted to relive meetings as audio. Reading a summary took 30 seconds; listening to AI-generated recap took 10 minutes.

TLDR taught me lesson one: Clever doesn't equal useful.

I pressed on. After taking two weeks off to backpack through Thailand, I came back with a refined philosophy. Instead of asking "What's the big market opportunity?" I started asking "What would our team use every day?" If the internal team adopts something naturally, it’s a strong signal that our community—an extension of that team—will too.

With that framework, I built Kairos.

Kairos was my attempt to make reading less solitary: an AI reading companion that let you highlight text and ask questions in real time. I built it in two weeks, wrote about it with staff writer Katie Parrott, and launched it in February. One thousand people installed it immediately.

Day one felt electric. Day three was silent.