Why we built Compose: an interview with Tjaž
I asked myself: what would I pay 30 euros a month for right now?
And the answer was obvious. Something that writes emails for me.
I spend 3-4 hours a week writing business emails. I'm good at it. But I don't love it. Each email is slightly different. Different tone. Different stakeholders. But the underlying work is the same: translate thoughts into professional language.
The Internal Tool
We built Compose in two weeks as an internal tool. It wasn't polished. It barely worked. But I could write an email in 30 seconds instead of 15 minutes.
Aljaž tried it and immediately asked why we hadn't shipped it to customers.
"Because it's janky?" I said.
"No, because it solves a real problem. Everything else is details."
He was right.
Finding Product-Market Fit
The first version failed. We positioned it as "AI email writing." Nobody cared.
We repositioned it as "never write a business email from scratch again." Suddenly people got it.
Usage doubled.
We added reports. Then proposals. Then document templates.
Usage doubled again.
The Insight
What we learned: people don't want AI. They want time back.
The moment we stopped talking about how smart our AI was and started talking about the 2 hours you'd save per week? Everything changed.
What's Next
We're working on two things:
- Tone preservation. The AI rewriting should sound like you. Not like a robot. Not like a copywriter. Like you on a good day.
- Collaborative drafting. Not everyone writes alone. How do you keep the AI input flowing when you're writing with a team?
These are the problems that'll define the next version.