Xavier Redó

Xavier Redó

Founder & CTO

33 articles by Xavier Redó

MarsBased logo above the Shopify bag, surrounded by the words Català, Euskara and Galego, with the caption "Now supported natively in Shopify

Shopify now speaks Catalan, Basque and Galician

Shopify adds official support for Catalan, Basque and Galician, extending translations beyond the theme to content such as the checkout.

Leer el artículo: Shopify now speaks Catalan, Basque and Galician
Three wooden arrows pointing in opposite directions on a beige surface

Decision models: practical AI where it used to be too costly

Decision models could make AI practical in applications where cost and complexity have kept it out. Here’s what they are and why we’re keeping an eye on them.

Leer el artículo: Decision models: practical AI where it used to be too costly
Claude Cowork

What I actually do with Claude Cowork

From Slack noise to curated briefings: How Claude Cowork transformed how I oversee projects and team alignment.

Leer el artículo: What I actually do with Claude Cowork
AI-Generated Code

How "Don’t repeat yourself" can mess up with your AI-generated code

AI agents often over-abstract coincidental duplication, creating fragile architectures that increase cognitive load and technical debt. Prioritizing clear domain boundaries over rigid DRY principles is essential for keeping codebases maintainable for both humans and AI.

Leer el artículo: How "Don’t repeat yourself" can mess up with your AI-generated code
Bye bye Cursor

The strategy behind the shift: Why MarsBased left Cursor for Claude Code

After starting with Cursor in 2024, we are now standardizing on Claude Code for our 2026 agentic strategy. This move marks our official shift from writing code to defining problems and steering the process.

Leer el artículo: The strategy behind the shift: Why MarsBased left Cursor for Claude Code
Upgraded legacy Rails app

How we safely upgraded a legacy Rails app with limited tests

Upgrading a 150k-line Rails monolith with low test coverage is risky. We avoided a "big bang" by using progressive rollouts and bot traffic to ensure a stable, incident-free migration.

Leer el artículo: How we safely upgraded a legacy Rails app with limited tests
Astronaut

How we made Plausible analytics work with ad blockers

To combat analytics data loss caused by ad-blockers, we implemented a custom Ruby on Rails proxy to route Plausible requests through our own backend. This privacy-friendly approach ensures we capture reliable traffic insights while fully respecting our visitors' data and privacy.

Leer el artículo: How we made Plausible analytics work with ad blockers
Context is everything with LLMs and AI assistants

Context is everything with LLMs and AI assistants

This might sound obvious to some, and maybe others are a bit late to the party, but one thing you have to take into account is this: context changes everything when working with LLMs and AI assistants. There is a massive difference between using ChatGPT, Gemini, Copilot, or any other LLM in isolati

Leer el artículo: Context is everything with LLMs and AI assistants
Docker

Always use a .dockerignore in your projects

A missing .dockerignore slows down deployments and risks exposing sensitive data. Learn why this small but critical file is as essential as .gitignore.

Leer el artículo: Always use a .dockerignore in your projects
Keyboard

AI code agents and Typed Languages

AI code agents work best with typed languages. Better inference, safer refactors, and clearer APIs make typing feel like leverage rather than bureaucracy.

Leer el artículo: AI code agents and Typed Languages
React

React vulnerability (CVE-2025-66478): How we responded and what you should do

A critical React SSR vulnerability (CVE-2025-66478, CVSS 10.0) allows remote code execution. We have swiftly patched all active React/Next.js client projects. Users must apply the official fix immediately.

Leer el artículo: React vulnerability (CVE-2025-66478): How we responded and what you should do
MCP Introduces a new standard for interactive AI interfaces

MCP Introduces a new standard for interactive AI interfaces

The Model Context Protocol (MCP) community has introduced a new proposal called “MCP Apps” that aims to standardize how interactive user interfaces are built and shown inside AI assistants.  This matters because interactive components already existed in practice, but there wasn’t a common MCP-level

Leer el artículo: MCP Introduces a new standard for interactive AI interfaces