https://www.lians.ai/blog Writing from the Lians team on regulated AI memory: bitemporal recall, audit chains, backtest contamination, and honest evals. en https://www.lians.ai/blog/bitemporal-ai-memory https://www.lians.ai/blog/bitemporal-ai-memory Sun, 19 Jul 2026 13:00:00 GMT A technical guide to point-in-time retrieval, late-arriving facts, corrections, and reproducible agent decisions. https://www.lians.ai/blog/ai-agent-audit-log https://www.lians.ai/blog/ai-agent-audit-log Sun, 19 Jul 2026 12:00:00 GMT A technical guide to AI agent decision records, memory provenance, bitemporal reconstruction, tamper evidence, and point-in-time replay. https://www.lians.ai/blog/eu-ai-act-article-12 https://www.lians.ai/blog/eu-ai-act-article-12 Fri, 17 Jul 2026 12:00:00 GMT A practical guide to Article 12 logging for high-risk AI systems and the evidence architecture agent teams should build now. https://www.lians.ai/blog/locomo-benchmark https://www.lians.ai/blog/locomo-benchmark Fri, 10 Jul 2026 12:00:00 GMT Lians reports 92.9% top-200 judged accuracy on LoCoMo versus Mem0's current 92.5%. The shared answer and judge harness, retrieval differences, artifacts, token costs, and caveats are published. https://www.lians.ai/blog/memory-lifecycle https://www.lians.ai/blog/memory-lifecycle Mon, 06 Jul 2026 12:00:00 GMT Lians v0.4.0: pre-compaction flush, conflict resurfacing, materiality-weighted decay, audited degraded retrieval, and a signed Markdown memory statement. Memory as a lifecycle, not a table. https://www.lians.ai/blog/backtest-lookahead https://www.lians.ai/blog/backtest-lookahead Sun, 05 Jul 2026 12:00:00 GMT Lookahead bias in AI agent memory silently inflates backtest results. FUTURE_EVENT is the obvious class; LATE_REVISION is the one vector stores can't even see. Here's the mechanism and the check.