We re preparing Kachilu for our March 2026 launch, and we want to stay focused on what actually helps teams ship.

Current priorities:

- no-setup onboarding

A lot of AI tools still assume single-player usage.

For Kachilu, we re focused on team operation from day one:

- secure, isolated workspace boundaries

We kept seeing the same issue across workflow tools: once business logic changes, static automations break and become costly to maintain.

With Kachilu, we are testing a different model:

- describe the goal in natural language

OpenClaw is hard for non-experts.

OpenClaw is trending, but in practice it s only usable by a limited group of people mainly because the initial environment setup is painful, and the configuration afterwards is even harder.

On top of that, I suspect many of OpenClaw s workflows will eventually be replaced by OpenAI s Codex.

But adopting Codex isn t easy either especially for Windows users. And most real work happens in teams. That s why I m convinced the truly exciting future is this:

I m currently building Kachilu, an AI-driven infrastructure platform powered by Codex.

One of the biggest challenges we faced was ensuring that AI-generated Terraform/HCL code is actually production-ready and secure, not just "syntactically correct." We ve spent months fine-tuning how Codex understands cloud dependencies.

I m curious: For those using AI to help with DevOps, what s the one thing you still don't trust the AI to do?

We re launching Kachilu on Product Hunt soon! If you re interested in the future of AI-managed infra, I d love to have you follow along for updates.