Oct 14, 2026 · 1 hr 21 min · 12 segments
Kai Hendry and Vincent De Smet debate build versus buy in the age of AI, infrastructure testing, agent workflows and the human bottlenecks in enterprise delivery.
So what do they do? Where did you go? We've made a Rust version of Git or something insane.
Oh, well, I think it's kind of in line with their previous bottleneck articles, as you called them.
Yes, they moved on from, I mean, it's in line of like we rebuilt CI, like they talked about CI as a product, meaning that it's not just GitHub Actions workflows, but it's, you know, actually a full-fledged like workflow engine that builds and, you know, they're doing.
I believe it's like an internal product, right? And Scion, the thing they built, is basically a monorepo version control system where they made a couple of challenges to the Git design.
I think one of the very interesting ones was that they don't want the client and server...
so the client doesn't need to carry all of the server capabilities, which allows them to rewrite the server internal logic without redeploying clients.
And then a lot of it is based on event orchestration and derived use and materialization of certain queries.
So clients only load what they need, And the server is basically providing a bunch of views that the clients can query.
Yeah, and also inside the blog post, there's some interesting information such as Spanner and cloud object storage or blob storage.
So what do they do? Where did you go? We've made a Rust version of Git or something insane.
Oh, well, I think it's kind of in line with their previous bottleneck articles, as you called them.
Yes, they moved on from, I mean, it's in line of like we rebuilt CI, like they talked about CI as a product, meaning that it's not just GitHub Actions workflows, but it's, you know, actually a full-fledged like workflow engine that builds and, you know, they're doing.
I believe it's like an internal product, right? And Scion, the thing they built, is basically a monorepo version control system where they made a couple of challenges to the Git design.
I think one of the very interesting ones was that they don't want the client and server...
so the client doesn't need to carry all of the server capabilities, which allows them to rewrite the server internal logic without redeploying clients.
And then a lot of it is based on event orchestration and derived use and materialization of certain queries.
So clients only load what they need, And the server is basically providing a bunch of views that the clients can query.
Yeah, and also inside the blog post, there's some interesting information such as Spanner and cloud object storage or blob storage.
The rest of this transcript — segmented and speaker-labeled, so you land on the exact moment something was said
Search every transcript — by keyword, by phrase, or by meaning, across every show Radar indexes
Trends — what is surging across podcasts, measured against its own baseline
Alerts — when a name you follow appears in a newly indexed episode
No account is needed to search Radar.