Brainfish Session Auto-Discovery and Sampling Behavior

Overview

This article describes how Brainfish automatically discovers sessions from recordings and creates knowledge in the CMS, including its default sampling behavior and performance/cost considerations.

Auto-Discovery of Sessions

Brainfish supports automatic discovery of sessions from recordings without requiring manual use of the Discover action.

When auto-discovery is enabled:

• Brainfish scans recordings to identify relevant sessions.
• Discovered sessions are automatically converted into knowledge entries in the CMS.
• Manual pressing of the Discover button on each recording is not required for those sampled sessions.

Sampling Behavior

Auto-discovery does not currently run on 100% of recordings. Instead, it uses sampling to balance coverage with performance and cost.

• The system samples approximately 5% of recordings by default.
• Only the sampled recordings are processed for auto-discovery and knowledge creation.
• The sampling rate is configurable by the Brainfish team and can be adjusted if required.

Performance and Cost Considerations

The sampling rate is intentionally conservative:

• Higher sampling percentages increase compute usage and associated costs.
• Running discovery on every recording may impact system performance, depending on volume.
• The default ~5% sampling is chosen to provide representative coverage while limiting impact on performance and cost.

If you need broader coverage (for example, more complete knowledge extraction from recordings), request an adjustment of the sampling rate, keeping in mind the trade-offs between:

• Coverage and completeness of discovered knowledge
• System performance
• Processing and infrastructure cost

Notes

• Auto-discovery is already live and actively creating knowledge entries in the CMS from sampled recordings.
• The sampling percentage is not user-configurable in the UI; changes must be made by the Brainfish team.