About Batru

What Batru is

Batru turns competitive-game data into decisions you can act on. Purpose-built models transform drafts, match data and changing metas into calibrated probabilities, recommendations and measured relationships — not vibes or stale lookup tables. Players use the web tools directly; AI assistants call the same live data through MCP. As Batru expands, each game gets its own model, methodology and product surface under one brand.

The name

Batru comes from baɣatur — the old steppe word for “champion” or “hero”. It travelled across Eurasia over centuries and survives today in names and titles from Central Asia to Eastern Europe (the Russian bogatyr, the Mongolian baatar in Ulaanbaatar, “red hero”).

We picked it because it captures the whole point of the product: become a champion. Better reads, better decisions, better wins — one match at a time.

How it works

Each game has its own model: a Transformer over the draft, trained from scratch, exported to ONNX and retrained as the meta shifts. We publish the training data, accuracy and calibration in full — no black boxes.

Built agent-native

Batru treats AI assistants as first-class users, not scrapers. Every number on this site — win probabilities, tier lists, counters, synergies — is equally available to LLM agents through our hosted MCP server and Python SDK, free and keyless. And because the probabilities are calibrated, they’re numbers a machine can act on directly — not vibes it has to second-guess.

Ask Claude, ChatGPT or Copilot about a draft and there’s a good chance the answer already cites batru.gg. Wire in the MCP server and it stops guessing entirely. Set it up in one line →