The facts behind the answers.

Every claim on this page is backed by a live API. No account needed, no rate limits. Copy any command below — or let your agent do it.

Seven years of structuring the world's reference data.

FYIPedia started in 2019, when we began building developer-facing data tools alongside fintech and infrastructure projects. The first was a color reference tool — because every existing converter was either buried in ads, couldn't handle modern color spaces like OKLCH, or had no API. We built ColorFYI, and that single decision — authoritative data, domain-specific models, zero-auth REST APIs — became the blueprint for everything after it.

Over the next five years, we refined the approach across increasingly complex domains. Chemistry required parsing IUPAC nomenclature rules. Ornithology demanded reconciling three competing taxonomies (IOC, Clements, and Howard–Moore) into a single coherent hierarchy. Sake brewing needed bilingual Japanese–English models that preserve 特定名称酒 classification. Cocktail chemistry meant computing ABV and calories from ingredient volumes, not copying numbers from a recipe book. Each domain taught us something new about data modeling, and each one proved the same thesis: people — and the agents they rely on — deserve reference data structured by those who understand the subject matter, not just the schema.

By 2024, we had systematized the process enough to accelerate: 14 sites by December, 30 by mid-2025, and 53 today — spanning 11 knowledge families from cocktail chemistry to stellar classification.

We don't dump data into generic tables.

Every FYIPedia site has a purpose-built data model designed from two perspectives: the data scientist who needs typed, relational, queryable fields — and the end user who needs computed values, intuitive groupings, and cross-references that make sense in context. We design schemas that encode domain expertise, not just store rows.

The pattern repeats across all 53 sites: foreign keys capture real-world relationships (a cocktail belongs to a family, an element sits in a period and group), computed fields derive values that users actually search for (ABV from ingredient volumes, wingspan from morphometric datasets), and every field traces to a specific source with version and date.

Every record traces back to a primary source.

We don't generate data, and we don't scrape aggregators. Each domain has a custom ETL pipeline that reads directly from authoritative datasets, validates against domain-specific schemas, and loads into typed PostgreSQL models.

Our 5-stage pipeline

Real data. Real sources. Ready for anything.

We maintain direct access to these authoritative datasets — not through aggregators, mirrors, or third-party APIs. Reference data you can verify: each source is cited on its FYIPedia site with version number and access date.

Data freshness

Not all reference data changes at the same rate. We match our update cadence to the nature of each dataset — from minute-level seismic feeds to annual standard releases.

Your AI already knows how to use our data.

Reference data is only useful if machines can access it reliably. We build native integrations for every major AI platform — so whether you're coding directly, delegating to an agent, or asking a question in natural language, you get structured data instead of hallucinated guesses.

From colors to cocktails to constellations.

53 sites organized into 11 thematic families. Each family has a hub domain, shared design language, and cross-linked navigation. Every member site shares the same API conventions, i18n framework, and data quality pipeline.

Use our data in any language.

Zero-dependency packages published on every major registry. Core computation packages (color math, unit conversion, distance calculation) work entirely offline — no API calls needed. API client packages wrap any FYIPedia site with full type annotations.

All packages MIT-licensed. Published under github.com/fyipedia. Rust crates and Ruby gems are code-complete, pending registry publication.

How we keep 53 sites reliable.

All sites share one proven stack — optimized for uptime, data integrity, and global performance rather than novelty.

How we ensure accuracy at scale.

With 800,000+ records across 53 domains, data accuracy is the product. We enforce it at every layer — from initial import through runtime serving.

Who builds FYIPedia.

Based in Seoul, South Korea and the United States, with contributors across Toronto, Berlin, and Singapore. A team of engineers and domain specialists who value subject-matter expertise as much as engineering skill — backgrounds in biochemistry, history of science, applied mathematics, industrial design, and marine biology. We understand the data we model, not just the code that serves it.

Plus 6 more specialists in marine biology (FishFYI), mixology & food science (Beverage FYI), aviation operations (Transport FYI), gemology & mineralogy (Science FYI), CJK linguistics (i18n), and DevOps/infrastructure.