advancespace

Author here. We build incident management software so obviously we're biased, but this post came from a pattern we keep seeing with prospects. Team builds an internal incident bot with AI in a weekend, everyone's impressed, then six months later someone's spending half their time on Slack API changes, on-call routing edge cases, and compliance work nobody scoped upfront. The initial build was never the expensive part.

This is the part people keep missing. The bottleneck was never typing code. It's understanding the problem deeply enough to know what the right solution looks like. AI made the cheap part cheaper.

This is an interesting framing but it assumes maintainers want to use agents at all. Most OSS maintainers we've talked to build things because they enjoy the craft. Donating compute to replace that craft is like offering a chef a microwave.

The review burden problem mirrors what happens with internal tools generally. Teams use AI to spin up an internal system in a weekend, everyone's impressed, then six months later someone's spending half their time maintaining it. The build was never the expensive part. The review, the edge cases, the ongoing maintenance - that's where the real cost lives, whether it's OSS contributions or internal tooling