Skip to main content
This agent is currently in preview and is not available to all customers at this time. Contact Sales for more information.
Encodes multivariate sensor data into fixed-size embeddings that capture how a system is behaving, the same representation behind every detection agent on the platform. Every sensor data time window comes back as one embedding vector per channel, ready for downstream ML.

Use Cases

  • Classify operating states from a few examples: Embeds labeled example windows and live windows into the same space, so a lightweight classifier works without training a model.
  • Score similarity across sessions and machines: Measures distance between embedded windows, so the same behavior is recognizable across runs, assets, and sites.
  • Explore data before committing to a pipeline: Projects embeddings into a map where operating states separate into visible clusters, so structure shows up before any modeling decision.