SES's low price can hide the surrounding infrastructure cost

Transcript

From the transcript Latent Space publishes with the episode.

1. 32:27

   Uh, what do you take from them, and then what do you-- what should people know to stay away from them? We started with SES too. Uh, we knew that, okay, if we're gonna be innovating here, I wanted to innovate on the DX layer, uh, because I felt like that was the underserved, uh, space. So let me have SDKs for every single language. I remember with Serverless, we highlighted service because, like, no one had, like, guides on how to send an email with Lambda or something. So there's just this underserved space in, in DX.

2. 32:59

   So we started with, like, the most high-level layer, and that, that served us really well. Was able to, like, scale because we were using SES behind the scenes. Turns out there's just a lot of limitations when it comes to the way you structure domains, and they have this concept of configuration sets, and the hidden cost if you're going with SES is that, okay, you might be paying only 5K, for example, but now you need SNS for the events.

3. 33:29

   So you need, like, all the feedback from bounces, complaints, open clicks, uh, all that. And now, so you get the events. Great. Now you need CloudWatch for the logs. Now you need a queue because it's gonna be a lot of events, so now you add SQS, and now that you have SQS, y- so you start, like, adding all these different layers. Oh, you want webhooks? Okay, now you gotta build another thing. So if you go with SES, it's a great platform.

4. 33:57

   It, it's fine, but you're gonna notice, like, your footprint will be much bigger than you would imagine very quickly because you need all the apparatus, uh, around it. Yeah. Um, I would say, you know, uh, another YC startup that came up in the mail area is AgentMail. Yep. And then they, they-- I talked to the guy as well, and he was like, "Yeah, we are not gonna-- we're gonna build on top of SES and not [laughs] really compete with them." Yeah, but for us, it's changing that now.

5. 34:22

   So we are-- we're building the infrastructure and going, like, down to the metal now. And for us, it, it, it feels like we got to the point where now this makes sense to innovate on the, the most basic layer, and now we can get better throughputs, better deliverability if we just focus on that. Uh, but it's a big project. Like, we've been working for months now, and we're gonna announce something soon about that route. What, what is the hardest thing to... Is it, like, uh, trusted IPs? Yeah.

6. 34:50

   Is it compute and data centers, or what, what is the blocker? Yeah, definitely IPs. You're buying, like, a whole IP block, and then now you're managing the IP pools yourself. That, that's one. And then you're almost doing, like, traffic control, uh, and you're controlling the throughput into each mailbox provider, and then you need the relationships with the providers. But then, like, once you have that established, that infrastructure, now you have way more control into what you can s- give to users. So one of the-- my biggest things were, like, observability, you know?

7. 35:25

   Like, that as a engineer, I, I wanna see exactly what's happening, and I've always felt like in this space there wasn't enough information. Like, once-- when an email goes out, is it really delivered? Is it really on the spam folder? Is it in the promotional tab? Like, I never know. Like, you just trust that the email goes out, and then you always hear from some user like, "Oh, I didn't get the email."

8. 35:47

   I'm like, "What?" Like, "What happened?" So there's that empty space between those things, and the more we're close to the metal, the more we have in terms of data that we can surface for, for users.