Building software lives or dies on the boring stuff.
Emails actually arriving in owners' inboxes. Push notifications firing when a request is logged. The system being up at 6am when the contractor signs in.
None of that is glamorous. None of it shows up in a demo. But it's the difference between software that helps a building run and software that becomes another thing the building manager has to chase or workaround.
A lot of that reliability isn't just our own code - it's the providers underneath us. Email infrastructure, push notification services, our database host, our CDN, our auth layer. Every one of those is a dependency, and every one of them can have a bad day.
Part of our job is choosing them carefully, monitoring them honestly, and having a plan for when one of them wobbles. We take that responsibility seriously because our customers can't tell the difference between "our software broke" and "something upstream broke" - and frankly, they shouldn't have to.
So we run on managed Postgres, sit behind Cloudflare, monitor deliverability obsessively, and we re-evaluate our stack regularly.
Not because it's exciting, but because a building waiting on a emergency broadcast doesn't care about our architecture diagram.
They care that the email arrived.