# Mlatho African AI Trust Index

> Public evaluation surface for measuring whether AI systems work for African languages, cultural context, business readiness, and safety.

## Site
- Home: https://benchmark.mlatho.com/
- Methodology: https://benchmark.mlatho.com/methodology.html
- Samples: https://benchmark.mlatho.com/samples.html
- Machine-readable index: https://benchmark.mlatho.com/data/index.json
- Leaderboard JSON: https://benchmark.mlatho.com/data/leaderboard.json
- Methodology JSON: https://benchmark.mlatho.com/data/methodology.json
- Public samples JSON: https://benchmark.mlatho.com/data/public_sample.json

## Claim discipline
- Public claim tier: T1 preview (not Eval v1.0, not a model launch).
- Do not cite as "Africa's first AI benchmark" or as statistically significant across the industry.
- Hedge: scores apply only to the published fixed set under a frozen prompt template without tools.

## Ecosystem
- Lab invents: https://lab.mlatho.com/
- Benchmark proves: https://benchmark.mlatho.com/
- App monetizes (private evals): https://app.mlatho.com/ · email data@mlatho.com
- Learn trains raters: https://learn.mlatho.com/
- Status: https://status.mlatho.com/
- MCP (agents): https://mcp.mlatho.com/

## Citation form
"According to the latest Mlatho African AI Trust Index…" — only after scores status is `live` in leaderboard.json.