Skip to content
Lucas Lessa
Back to projects

Case study · Product · AI applied to data

Café com Features

A portal that brings Databricks, Snowflake and Microsoft Fabric release notes together in one place, turned by AI into plain-language summaries with practical examples of how to use each feature.

A robot runs every day, collects the updates from the three platforms, fetches the related documentation and publishes the news organized by platform, month and release stage (GA, Public Preview, Beta).

  • Next.js
  • TypeScript
  • Supabase
  • Gemini
  • GitHub Actions
  • Vercel
  • Tailwind CSS

The problem

Keeping up with several data platforms takes time every single day:

  • Each platform publishes release notes in a different format and place.
  • Notes are short and technical: they say what changed, but rarely how to use it.
  • It is hard to tell what is already GA from what is still Preview or Beta.

How it works

  1. Daily collection

    A GitHub Actions job runs every day at 5 a.m. and reads the official sources: the Databricks RSS feed, the Snowflake release notes pages and the Microsoft Fabric changelog on GitHub.

  2. Only what is new

    Each note is checked against the database by its source id; only new ones move on.

  3. Documentation context

    The robot opens the documentation page linked to the note to get details and code examples.

  4. AI summary

    Gemini (free tier) writes, in Portuguese and English, a structured text: what it is for, how to use it, an example, limitations, stage and impact.

  5. Publishing

    News items go to Supabase and the Vercel site is refreshed right away (on-demand revalidation).

Main features

  • Three platforms

    Databricks, Snowflake and Microsoft Fabric in the same feed.

  • Filters

    By platform, month and release stage (GA, Public Preview/Beta, updates), plus search.

  • Plain-language explanations

    Each item has practical examples and how to apply the feature in a pipeline.

  • Bilingual

    All content in Portuguese and English.

Architecture

LayerTechnology
WebsiteNext.js (App Router), TypeScript and Tailwind CSS on Vercel
DatabaseSupabase (Postgres with Row Level Security)
Ingestion robotNode.js scheduled on GitHub Actions
AIGemini API (free tier) with fallback models

Zero running cost: everything runs on the free tiers of Vercel, Supabase, GitHub Actions and Gemini.