Aadil Mamujee

My biggest pet peeve at work is people not doing something when they said they would. I’ve spent years tracking other people’s commitments alongside my own — what they said they would do by when. Many of those commitments get dropped the moment they’re made. Someone says “I’ll send you that doc by Monday” in a DM, both people agree, but it’s never tracked. I’m guilty of it too.

In 2021, I wrote a spec for a Slackbot called HandShake to solve this problem for myself. A bot sitting inside my DMs, recording the commitment, would catch the dropped balls before either of us did.

I tried to build it. I got a half-finished shell working, and gave up because it was hard and I didn’t have the skills. Slack apps in 2021 meant a backend, persistence, OAuth, and a thousand small product decisions I couldn’t make without writing most of the code first. I’m a product person who can read code and write small scripts, but it didn’t make sense for me to learn to build something I wasn’t even sure anyone but me would use. So the shell sat in a private repo, and the spec sat in Notion, both untouched for five years.

In the last few weeks, I built it — makepact.co. The Slack app, the landing page, the Stripe integration — all of it came together in a couple of weeks using Polsia, an AI agent platform. The 2021 spec barely changed and months of engineering work collapsed into a few weeks of product decisions.

The first thing Polsia did was spin up a landing page within minutes of dropping in the spec. Going from a paragraph in a Notion file to a complete landing page in a few minutes was the moment Pact stopped feeling like a dormant spec and started feeling like a product I could build. There’s a Stripe upgrade path, a Slack directory listing, a few real workspaces with the bot installed, a handful of pacts completed. This is not a real company, and I never intended it to be a company, but most of my side projects are not nearly this complete.

The problem is that Pact only works if other people use it with you. A pact between me and nobody is just a to-do list with extra steps. The real work isn’t building the bot anymore; it’s sitting with teams, watching how they use it, fixing the friction, and repeating that loop for months. That’s probably the right way to build a company. It’s also not what I set out to learn. I wanted to get better at building with AI, not spend my limited free time doing enterprise customer development between school drop-offs and time with my baby.

So Pact will die. The shell I half-built in 2021 died at the spec. This version dies on a real domain, with a working app, real Stripe payments, a real Slack listing, and zero traction. Five years ago I couldn’t build this at all. Now I can. That still feels like progress.