Key takeaway: Saving 2,000 hours with AI email orchestration

Like, can't put AI on bad data Right.

Or you're gonna have a bad time. Yeah. So the first step was zoom out, get the data in a place that's clean and readable, and then also in a place where a computer can do the compute. Yeah. Like the rules we talked about, all those need certain values that the computer's expecting to calculate this and do it all automatically.

And then as you see on the blueprint, there's a lot of stuff in people's heads. So we had a full time coordinator. It was just in her head. She's wonderful.

She's great. But you just can't scale that way. We do the rules which is over 30 rules. Some of them we talked about.

And then the audience overlap is huge. Again, that was a big component for us. Yeah. And that's handled in Databricks where we look at all the audiences.

We look at how they overlap by day by send. It's like 18,000 audience combinations that we just couldn't do with a human.

And then out of all that, it tells us the best date time for the email.

Wow. 18,000 calculations. That's... Yeah. No human can do that. You'd have to be a superhuman. I really like Genie.

We probably couldn't have done this project without it. If I estimated, it probably did about eighty hours of work I didn't have to do of dev work. I'm like, Jeanie, do this. So it's a really good junior developer, I guess. Yeah. But if you need a content writer, you could lean into that. For us, it took us, I wanna say about a month, four weeks.

But without AI in a... If this was two years ago, this would have been a team of eight and it would have been several months and then hopefully it worked. This was just transformational for us and I hope you all can do something similar.