To get a sense of how well the engine handles a real project, we asked it to build a Calendly-style booking system from a single prompt. It ran the whole process — planning, code generation, tests, infra, and deployment — in about 8 hours.

What it generated: - ~20K lines of Python (FastAPI, async)

- Postgres schema (6 tables)

- Services, background tasks, booking logic

- Email notifications

- 40 unit tests + 22 integration tests

- Docker Compose (API + Postgres)

- GitHub Actions pipeline

- A running deployment tied to Postgres

- Code & live endpoints

Everything is open: Repo: https://github.com/OlivierKodaii/calendarKodaii

API docs: https://calendar.kodaii.dev/docs

OpenAPI schema: https://calendar.kodaii.dev/openapi.json

Admin interface: https://calendar.kodaii.dev/admin

Why we’re sharing this

We think this line of work may be of interest to peers on HN who care about backend architecture, tooling, and large-scale code generation and your feedback would be very much appreciate.

If you inspect the repo, we’d appreciate any comments on:

- structure,

- code clarity,

- decisions that look odd,

- failure points,

- or anything that feels promising or problematic.

For those who want to follow the project or join the upcoming alpha: https://kodaii.dev

Happy to discuss details or answer questions.