Documentation
Install the Cloudmind SDK, send your first trace, and set up evaluations. Reference and guides for the whole platform.
Cloudmind traces every model call your application makes, scores the outputs against evaluations you define, and keeps your prompts under version control.
If you are starting from scratch, work through Getting started in order. It takes about fifteen minutes and ends with traces flowing and one evaluation running.
The mental model
Everything in Cloudmind hangs off a trace: one top-level operation in your application, containing however many model calls, tool calls and retrieval steps it took.
- A trace records which prompt revision produced it.
- Interesting traces get promoted into a dataset.
- Evaluations run against datasets and produce scores.
- Monitors alert when scores, cost or latency drift.
Read Core concepts for the detail.
Where to go next
Pick the path that matches what you are doing today.