Questions, answered.

The map of human history — a world map you can rewind. A 3D globe with a single time thread: drag the year and borders move, kingdoms rise and dissolve, monuments appear when they were built. An AI historian answers questions about any place and era you select, from sourced, confidence-tagged facts. The name is Sanskrit: सत्य means truth.

Chapter one is India, 1526–1947 — Babur's victory at Panipat to Independence and Partition: 21 powers with real boundary snapshots (or honest capital points), 123 dated rulers and officeholders, 47 located events, and 21 landmarks that appear in the years they were built. India is the first chapter, not the whole product — the engine is civilization-agnostic and more chapters are planned.

Boundaries come from the historical-basemaps scholarly compilation — real snapshot years, simplified, labelled approximate by design. Rulers and events are editor-curated from standard reference histories and cross-checked against encyclopedic sources, with deep links for further reading. Every data point carries a source reference, and the interface shows sources alongside claims.

No — by design. Satya's AI answers only from the source-linked facts inside your selection of place and time. It states attested facts plainly, hedges probable ones, flags disputed ones, and calls legends legends. When no sources exist for a selection, it says that nothing is recorded there in the current dataset — instead of guessing. No source, no story.

Two independent tags on every claim: temporal precision (exact, circa, estimated…) for how well the date is known, and credibility (attested, probable, disputed, legendary) for how well the claim itself is supported. "When it happened" and "whether it happened" are different questions, so they're tagged separately.

No map of this era can honestly claim exact borders, so Satya doesn't. Frontiers were zones, not lines. Satya uses real scholarly snapshot boundaries, morphs between them so you can feel change over time, and labels the result approximate by design. Powers with no reliably mapped territory appear as capital points rather than invented polygons.

Satya's core gesture: draw a rectangle on the map, set an era on the time thread, and ask. Your selection in space × time becomes the question, and Satya answers from the sourced facts inside it — the kingdoms holding that ground, the rulers reigning, the events that happened there, and the monuments standing at that year.

That's the design goal — Satya serves two audiences: people who explore the atlas, and machines that need history to come back cited instead of hallucinated. The underlying data model (sourced, confidence-tagged assertions) exists today; a public, versioned query API for agents is on the roadmap and not yet available.

The atlas runs in a modern desktop browser with JavaScript enabled — no account needed. The globe, time thread, kingdoms, events, landmarks, and the lens's local sourced summaries all work as-is. The AI historian's long-form voice requires a server-side API key; without one, Satya says so honestly and gives you its local sourced summary instead of faking an answer.

Because it wasn't. At any year on the map, many powers, peoples, and versions of events coexist — 1565 reads differently from Vijayanagara than from the sultanates that defeated it. Satya's thesis is that truth (सत्य) means refusing to fake certainty: show every side, show the sources, show the doubt.