1
0 Comments

The complete tech stack of Cartesiano.ai

Here is a complete breakdown of everything used to create this Saas, it's cost, and why I used it:

Development

  • Ruby on Rails + PostgreSQL ($0): proven 'boring' technology to build out the user-facing admin, with lots of easily-available gems to simplify development of non-trivial flows (i.e. security, user management, etc.).

  • Python + FastAPI ($0): for LLM processor (the service which connects to multiple LLMs and processes their output).

  • VSCode + Github Copilot Pro ($10/month): Copilot Pro gives me access to coding assistants

  • Claude ($5/month): so far, this has been enough credit for my development needs, using mostly Claude Sonnet 4.5

Hosting

  • Hetzner with 2 VPS ($9/month)

    • 1 VPS for Rails + Python app: both running via Docker

    • 1 VPS for PostgreSQL

  • Wordpress ($26/year): for hosting the blog

  • Mintlify ($0/month): hosts our external product documentation

Product / Operations

  • Betterstack ($0): log management & observability into how the apps are performing

  • Posthog ($0): for product analytics

  • Resend ($0): for sending emails

My learnings

In the past, I've made the mistake of spending way too much time thinking about the best tools and processes for hosting, scaling, etc. I've learned the hard way that none of this makes sense unless you have real users demanding 99.999% uptime. For this reason, I've opted for simplicity across everything, and using as much as I can free tiers.

posted toAvatar for product Cartesiano.ai
Cartesiano.ai