大多数“与数据聊天”工具仍是黑箱,用户无法验证答案。Mora反其道而行:将问题映射到语义层,直接对实际schema运行SQL,并在面板中展示查询语句与底层数据行,用户可自行验证或修正。通过添加MCP层,同一引擎可在Claude和Cursor中运行,使数据分析变得可治理且可信。引用@pladevall称,分析是AI最大机会,目前工具远未满足需求,因此今天正式发布Mora。
Most "chat with your data" tools are still black boxes. You ask a question, you get a number, and you have no real way to know whether the model invented it.
Mora goes the other direction. It maps the question to a semantic layer, runs SQL against your actual schema, and puts the query and the underlying rows right there in the panel. You can verify every number yourself, or fix the join if it got it wrong.
Add an MCP layer so the same engine runs inside Claude and Cursor, and this stops looking like another AI demo and starts looking like governed self-serve that data teams might actually trust. Worth a look if you are tired of being the bottleneck for every revenue question.