I'm building an AI automation agency in public. Here's what the first few months actually looked like.
Not the highlight reel. The actual thing.
Six months ago I started taking on clients for AI automation work. No big launch, no viral moment, just cold outreach and a lot of "we'll think about it" replies. Took me longer than I expected to land the first few... and even longer to figure out what clients actually needed versus what I thought they needed.
Here's what I've built so far and what happened after.
A dental clinic was sitting on a dead leads list, people who had enquired but never booked. Nobody was following up. I built a reactivation workflow that messaged them automatically over a few days with the right context. Recovered $18K in booked appointments in the first run. The owner called me confused, like something had gone wrong, because it worked too fast.
Another client was losing 20 hours a week across three manual processes. Lead responses going out late, reports being built by hand, follow-ups falling through. I connected the workflows they already had, automated the repeating parts, and within 90 days they'd recovered $120K in operational costs. The fix itself was embarrassingly simple.
Turns out the hardest part of this business isn't building automations. It's convincing people that simple, consistent systems beat complicated ones they'll never actually use.
I'm going to keep sharing these openly... the wins, the workflows, the things that didn't work. If you're building something similar or just curious how this kind of work actually gets done, follow along.
And if your business has workflows that are quietly eating time and money, happy to take a look.