I’ve been a CTO for the last few years, and lately I've been watching the whole "vibe coding" trend closely as more non-technical people build apps. It’s amazing and scary what non-technical teams can generate in an hour, but the magic breaks down the second they try to actually ship it.
They get stuck in "localhost" and end up tossing this AI-generated code over the fence to engineering. Now the eng team has to decide if it's worth allocating time to this problem and if so, start provisioning, securing, and maintaining this orphan code.
To fix this, I'm building g13u. It acts as an automated SRE agent for AI apps. You point it at a repo, and it auto-provisions the DBs, sets up auth, handles schema migrations, and patches vulnerabilities. Basically, it's a managed runtime so these teams can actually ship without bothering engineering. Deployed in your own cloud.
Since I've lived the vendor lock-in nightmare, I built in an "eject button." Everything is backed by standard Terraform, so you can export your state and deploy right into your own AWS/GCP whenever you want.
I'm heads down building this right now and would love some brutal feedback from this crowd. Specifically:
- Are you seeing this problem in your orgs?
- What's the most painful Day 2 maintenance task you'd want an agent to handle first?
I'll be hanging out in the comments all day. Let me know what you think.
Hi HN,
I’ve been a CTO for the last few years, and lately I've been watching the whole "vibe coding" trend closely as more non-technical people build apps. It’s amazing and scary what non-technical teams can generate in an hour, but the magic breaks down the second they try to actually ship it.
They get stuck in "localhost" and end up tossing this AI-generated code over the fence to engineering. Now the eng team has to decide if it's worth allocating time to this problem and if so, start provisioning, securing, and maintaining this orphan code.
To fix this, I'm building g13u. It acts as an automated SRE agent for AI apps. You point it at a repo, and it auto-provisions the DBs, sets up auth, handles schema migrations, and patches vulnerabilities. Basically, it's a managed runtime so these teams can actually ship without bothering engineering. Deployed in your own cloud.
Since I've lived the vendor lock-in nightmare, I built in an "eject button." Everything is backed by standard Terraform, so you can export your state and deploy right into your own AWS/GCP whenever you want.
I'm heads down building this right now and would love some brutal feedback from this crowd. Specifically:
- Are you seeing this problem in your orgs? - What's the most painful Day 2 maintenance task you'd want an agent to handle first?
I'll be hanging out in the comments all day. Let me know what you think.