Multitemporality

Every entity in HASH is versioned along two independent axes of time. Transaction time records when the system learned something, while decision time records when that fact was actually true in the world. Together they let you ask not just ‘what is true?’ but ‘what did we believe was true, and when?’

Travel to any moment and query the graph exactly as it existed then. Reconstruct the state of your data at the close of last quarter, replay how a situation unfolded, or compare what you knew at the time a decision was made against what you know now — all without maintaining brittle snapshots or audit tables by hand.

Updates never destroy what came before. Instead of overwriting a value, HASH appends a new version and preserves the full history of every entity and relationship. The complete lineage of each change stays available as first-class provenance, so the past is always recoverable and auditable.

Bitemporal versioning turns time into something you can reason over. Correct a historical record without losing the original, distinguish a late-arriving fact from a real-world change, and give agents the temporal context they need to draw causal conclusions rather than working from a flattened, present-only view.