The problem with existing no code tools is that you still need to spend a lot of time building after figuring out what is needed (a problem of its own). Being able to focus on what is needed means you can build faster than designing, dragging, dropping, configuring.

What can you describe? Everything from tables, field types, relationships between tables and columns/fields, role-based access restrictions (RBAC), reports, dashboard metrics, bar/line/pie charts, lists.

This is made possible by fine-tuning Llama 3.1 to convert requirements into specs and code. Applications evolve over time and this means that the model has to be predictable between versions to ensure that the data/schema migrations can be generated cleanly, as well as provide any errors and warnings related to the requirements or data loss.

For non-developers this can be a huge time-saver since you don't have to use any additional developer resources. For developers it can be a great way to get 80+% of the way for complex apps (it's like super-scaffolding). Generated apps can be fully evaluated before purchase. A full git repo containing the back-end, front-end, migrations, etc. is the output of the entire process.

Currently the generated code base is TypeScript, JavaScript, React, Next.js but the capability is there to generate different stacks because of the approach taken to not go directly to code when fine-tuning.

Brought to you by the team at Red Axle (https://redaxle.com)

Check out the video at https://youtu.be/rPppwB4jPlQ for a smaller app.

We look forward to your thoughts. Does building apps this way make sense for the super savvy HN crowd?