Tag
A new hosted API memory layer for AI agents returns a proof tree with every answer, including bi-temporal versioning, audit trails, and hash verification, achieving 80.2% on LongMemEval-S with transparent benchmarks.