Kavara™

Kavara™ is an AI-as-a-Service platform built around Kirk, its discovery engine. Kirk helps organizations explore datasets to reveal phenomena already present in the data—such as changing relationships, recurring behaviours and unexpected structure—which people and agents can then investigate to understand their meaning.

The platform supports seven data shapes: measurements over time, timestamped events, relationships and networks, measurements across locations, repeated experiments or operating cycles, numerical representations of text/images/audio, and unordered samples or static tables. Kirk maintains a catalogue of 32 categories spanning financial markets, cybersecurity, manufacturing, weather systems, human physiology, genomics, and other domains.

Discovery follows a workflow: first establish data fit and supported representation, then use Kirk to make phenomena visible, and finally investigate possible explanations. Kavara™ emphasizes the separation between discovering a phenomenon and explaining its cause, and provides tools to keep execution accountable through documentation of model configuration, runtime context and reproducible runs.

The platform offers API access for developers, agent integration via a developer guide on GitHub, and deployments that can use confidential compute (Intel TDX, AWS Nitro Enclaves, AMD SEV) to keep customer data cryptographically isolated from underlying infrastructure providers.