Mundevo
Carlos Fraile, builder of Mundevo

Author

Carlos Fraile

Builder of Mundevo. Background in software engineering and product, currently focused on personal-finance decision tools and the data systems behind them.

Why Mundevo

Mundevo started as a personal frustration: most cost-of-living tools answer the wrong question. They tell you a city's index value without translating it into what you actually need to negotiate, save, or budget. The site is the answer to that — what gross salary do I need to live there, with my lifestyle, accounting for taxes and what my employer would actually pay me?

I build it because relocation decisions are dense and high-stakes. A clean, honest decision-support tool removes a lot of friction for people considering where to live next.

How I work on this

  • Hand-curated where it matters. 31 of 35 cities have data cross-referenced against multiple public sources (Numbeo, OECD, PwC, Speedtest, WHO). The remaining 4 carry explicit AI-estimated disclaimers because their underlying volatility makes a single curated figure misleading.
  • Transparent methodology. Every page surfaces its own methodology block. The full calculation rules are at /methodology and the underlying public sources at /sources.
  • Build in public. The site's quarterly review cycle is documented on /about. Corrections from readers with on-the-ground knowledge are the fastest path to improving specific data points.
  • Editorial commentary is AI-assisted, then cached. Each city's short "Analyst take" is generated by Claude (Anthropic) from the structured data, then deterministically cached — same city + lifestyle returns the same prose across rebuilds. The bulk of the content (visa guides, methodology, glossary, case studies) is hand-written.

Where to find me

Editorial principles

A few constraints that shape how content goes up on Mundevo:

  • No fabricated specifics. Visa pages don't publish income thresholds (those change). Salary pages publish bands, not negotiation points. Cost figures cite sources and date-stamp themselves.
  • No SEO content for SEO's sake. Every page has a real use case behind it. The 1300+ URLs aren't auto-generated padding — they're combinations of the same dataset surfaced at different angles (city × lifestyle, city × country, role × city) so readers can find the framing that matches their question.
  • Affiliate links don't bend editorial. Partner selection per page is algorithm-driven (matching specialty to page intent). What the editorial says about each partner is independent of what they pay.