---
title: "Brainfish Auto-Discovery of Sessions and Knowledge Creation"
description: "Overview"
canonical_url: "https://help.brainfi.sh/articles/brainfish-auto-discovery-of-sessions-and-knowledge-creation-IguDd9i0SM"
md_url: "https://help.brainfi.sh/articles/brainfish-auto-discovery-of-sessions-and-knowledge-creation-IguDd9i0SM.md"
---
# Brainfish Auto-Discovery of Sessions and Knowledge Creation

## Overview

This article describes how Brainfish automatically discovers sessions (e.g., call recordings, session replays, or Slack conversations) and creates knowledge entries in the CMS, including how sampling works and key considerations for performance and cost.

## Auto-Discovery and Knowledge Creation

Brainfish supports automatic discovery of sessions without requiring a manual **Discover** action on each recording. When enabled:

* Brainfish scans a subset of available sessions.
* It automatically generates knowledge from those sessions.
* The generated knowledge is stored directly in the CMS (as drafts or suggested articles, depending on your workspace configuration).

This removes the need to manually press **Discover** on individual recordings to create knowledge, although you can still use manual Discover for specific high‑value sessions when needed.

## Sampling Behavior

Auto-discovery operates on a sampling basis rather than processing every single session.

* Default sampling rate: approximately **5%** of sessions.
* The sampling rate is **configurable** and can be adjusted by the Brainfish team if needed for your workspace.

### Rationale for Sampling

Sampling is used to balance:

* **Performance**: Avoids heavy processing load on large volumes of recordings.
* **Cost**: Reduces compute and storage costs by not processing every session.

If your use case requires a higher or lower sampling rate (for example, regulated industries, limited volume but high value per interaction, or intensive QA workflows), contact the Brainfish team to update the configuration.

## How Auto-Discovery Relates to Manual Discover

* **With auto-discovery enabled**: Brainfish continuously selects a sample of sessions and processes them into knowledge suggestions automatically.
* **Manual Discover**: You can still click **Discover** on any specific session or recording that you know is high-value (e.g., a key support call, an important bug report, or a rich product walkthrough). These manually submitted sessions are queued for analysis in addition to the auto-sampled ones.
* Auto-discovery does **not** prevent or override manual Discover; they work together to keep your knowledge base growing from both ad-hoc and systematic inputs.

## Slack Agent Context (Optional)

In Slack channels, Brainfish may post automated system messages if the channel is not yet connected to a Slack Agent. To use Brainfish fully in a Slack channel and allow knowledge creation from conversations:


1. Create a **Slack Agent** in Brainfish.
2. Add the relevant Slack channel in the **Connection** section of the Agent.

Once configured, Brainfish can:

* Participate in the channel through the Agent (e.g., answer questions via `@Brainfish` or slash commands, depending on your setup).
* Automatically discover and create knowledge from relevant conversations, according to the configured sampling settings.

## Key Points

* Auto-discovery of sessions is **live and available**.
* Manual pressing of **Discover** on each recording is **not required** when auto-discovery is enabled.
* Sampling is set to **\~5% by default**, adjustable by Brainfish for your workspace.
* Sampling levels are chosen with **performance and cost** in mind.
* Auto-discovery applies to recorded sessions such as call recordings, session replays, or Slack conversations, depending on your enabled integrations.
* Ensure a **Slack Agent** is configured if you want Brainfish to operate within a specific Slack channel and use those conversations for knowledge creation.