Adam Jacob

Very big one was we were trying to discover how to build this reactive engine that would be able to scale to really large, to conceptually scale to really large graphs and really large infrastructures and applications. And so we built that engine a couple of different times.

Very big one was we were trying to discover how to build this reactive engine that would be able to scale to really large, to conceptually scale to really large graphs and really large infrastructures and applications. And so we built that engine a couple of different times.

And one of the passes of it, the most recent, not the most recent pass, but the pass right before, used a lot of very old techniques. It looked a lot like an Oracle application from the 90s. A lot of stored procedures and a lot of like smart activity in the database because we're basically building a custom database engine.

And one of the passes of it, the most recent, not the most recent pass, but the pass right before, used a lot of very old techniques. It looked a lot like an Oracle application from the 90s. A lot of stored procedures and a lot of like smart activity in the database because we're basically building a custom database engine.

And one of the passes of it, the most recent, not the most recent pass, but the pass right before, used a lot of very old techniques. It looked a lot like an Oracle application from the 90s. A lot of stored procedures and a lot of like smart activity in the database because we're basically building a custom database engine.

So we embraced that and built a custom database engine inside of Postgres, which on some levels was magnificent. But in the end, it got overwhelmed by the things that always overwhelmed those systems, which is it's a little complicated to deal with. The table structure gets complicated. More complicated performance becomes a problem, not that those are unfixable.

So we embraced that and built a custom database engine inside of Postgres, which on some levels was magnificent. But in the end, it got overwhelmed by the things that always overwhelmed those systems, which is it's a little complicated to deal with. The table structure gets complicated. More complicated performance becomes a problem, not that those are unfixable.

So we embraced that and built a custom database engine inside of Postgres, which on some levels was magnificent. But in the end, it got overwhelmed by the things that always overwhelmed those systems, which is it's a little complicated to deal with. The table structure gets complicated. More complicated performance becomes a problem, not that those are unfixable.

And so then you had to back up and try again with a completely different architecture, with a very different approach to how everything would work. The important piece of that was we had a moment where we believed that that old engine was going to be good enough to get a couple of customers to experience the product while we were working on this newer engine that we knew would scale better.

And so then you had to back up and try again with a completely different architecture, with a very different approach to how everything would work. The important piece of that was we had a moment where we believed that that old engine was going to be good enough to get a couple of customers to experience the product while we were working on this newer engine that we knew would scale better.

And so then you had to back up and try again with a completely different architecture, with a very different approach to how everything would work. The important piece of that was we had a moment where we believed that that old engine was going to be good enough to get a couple of customers to experience the product while we were working on this newer engine that we knew would scale better.

So I split the baby in terms of having the team work on both. So there was a set of people who were... optimizing the old engine, just trying to get it good enough. And another team that was working on what was obviously going to be much better long term and even near term, but was going to take a little longer to build.

So I split the baby in terms of having the team work on both. So there was a set of people who were... optimizing the old engine, just trying to get it good enough. And another team that was working on what was obviously going to be much better long term and even near term, but was going to take a little longer to build.

So I split the baby in terms of having the team work on both. So there was a set of people who were... optimizing the old engine, just trying to get it good enough. And another team that was working on what was obviously going to be much better long term and even near term, but was going to take a little longer to build.

And while I'm glad that we did it because it did allow us to get a little more directed customer feedback, the customer feedback we learned was that it was too slow. We had to just come back to the team and be like, hey, we onboarded this customer. It quickly became too slow to use. So we're not doing that anymore. And it's all hands on deck. Everybody go work on the new engine.

And while I'm glad that we did it because it did allow us to get a little more directed customer feedback, the customer feedback we learned was that it was too slow. We had to just come back to the team and be like, hey, we onboarded this customer. It quickly became too slow to use. So we're not doing that anymore. And it's all hands on deck. Everybody go work on the new engine.

And while I'm glad that we did it because it did allow us to get a little more directed customer feedback, the customer feedback we learned was that it was too slow. We had to just come back to the team and be like, hey, we onboarded this customer. It quickly became too slow to use. So we're not doing that anymore. And it's all hands on deck. Everybody go work on the new engine.

And that's the way it is. As much as it's about the story of sort of the technical mistake or the failure, I think a lot of those stories really need to be about, like, how quickly did you learn that it was a failure? And then how quickly did you adjust? And how much time did you spend crying over the metaphorical spilled milk?

And that's the way it is. As much as it's about the story of sort of the technical mistake or the failure, I think a lot of those stories really need to be about, like, how quickly did you learn that it was a failure? And then how quickly did you adjust? And how much time did you spend crying over the metaphorical spilled milk?

And that's the way it is. As much as it's about the story of sort of the technical mistake or the failure, I think a lot of those stories really need to be about, like, how quickly did you learn that it was a failure? And then how quickly did you adjust? And how much time did you spend crying over the metaphorical spilled milk?