Vector search rollback plans matter more than the first demo

I worked on a product search feature where the first vector demo looked great. A few test queries found the right catalog items, and the team wanted to push it live fast. The issue appeared after a merchandiser renamed a product family and moved several SKUs into a new category. The text in the database changed, but the embedding index still had the old product language, so search results felt…

Related public posts

  1. 向量搜索今天召回突然变少,我先查 embedding 还是过滤条件 tech-data-ai · rant · 5 replies 2026-06-17T13:40:36.956Z
  2. Cursor 生成的代码总是改乱项目?后来我发现问题根本不在 AI tech-data-ai · rant · 1 replies 2026-06-08T18:07:17.427Z
  3. 数据分析转AI工程师需要补哪些技能 tech-data-ai · rant · 2 replies 2026-06-04T13:56:59.249Z
  4. LLM API cost monitoring best practices tech-data-ai · rant · 3 replies 2026-06-05T13:28:56.328Z
  5. How to evaluate RAG answers before putting them in production tech-data-ai · rant · 1 replies 2026-06-04T17:51:10.678Z
  6. Cursor 安装完成后一直无法连接 AI?我是这样排查解决的 tech-data-ai · rant 2026-06-08T18:02:22.461Z
  7. pgvector和Milvus怎么选,做向量检索别只看性能 tech-data-ai · rant 2026-06-06T13:07:51.294Z
  8. AI 模型效果突然变差,我先查特征漂移还是提示词 tech-data-ai · experience · 7 replies 2026-06-15T14:30:48.699Z
  9. What I learned fixing duplicate embeddings in a product search index tech-data-ai · experience · 5 replies 2026-06-15T05:18:21.815Z
  10. AI 标注结果忽高忽低该先查什么 tech-data-ai · experience · 2 replies 2026-06-13T20:19:02.520Z