Retrieval recall on LongMemEval-S: 99.2% R@5, and how we measured it — Galdo
Galdo's hybrid retrieval puts the right evidence in the top 5 for 99.2% of LongMemEval-S questions — no reranker, no LLM in the loop. Full methodology, per-category breakdown, a depth-sensitivity sweep proving it isn't a candidate-pool artifact, and the limitations we're not hiding.
About Galdo
Galdo shows you how your company actually works. It learns how your best people work, as the work happens, and keeps the decisions, exceptions, and lessons behind the work, including how they change. People ask in Slack; agents ask through the API, SDKs, or MCP, and answers come back with the source. US-hosted, with a free tier of 1K writes and 10K searches per month.