Fishladder

AI Analysis

Fishladder uses AI to help you extract insights from unstructured feedback. These features require the Anthropic API key to be configured by your workspace admin.

Transcript Analysis

Paste a customer call transcript, support conversation, or meeting notes into the analysis panel. The AI extracts:

  • Key feature requests and pain points
  • Suggested feedback entries with titles and descriptions
  • Impact and urgency assessments

Review the suggestions and click Create to add them as feedback entries in one step.

Auto-Grouping

When you have many ungrouped feedback entries, the auto-group feature analyzes all entries and suggests thematic groups. It looks for common patterns in titles and descriptions to cluster related requests together.

  1. Click Auto-Group in the feedback toolbar.
  2. Review the suggested groups and assignments.
  3. Accept or modify the suggestions before applying.

AI Feedback Ingestion

Beyond manual transcript analysis, the AI can automatically process incoming feedback. When feedback arrives (via email, API, or the UI), the AI ingestion pipeline:

  1. Analyzes the text to extract key themes and sentiment
  2. Checks for duplicates against existing feedback entries
  3. Creates or proposes new entries based on your permission settings
  4. Records signals for recurring patterns

This is rate-limited to 20 ingestions per hour per team. Text between 50 and 100,000 characters is accepted.

Deduplication

Before creating a new feedback entry, the AI compares it against existing entries to detect duplicates. If a likely match is found, it merges the data rather than creating a duplicate. This keeps your feedback board clean and consolidates related reports automatically.

Signal Detection

As the AI processes feedback, it detects recurring patterns and logs them as signals. For example, if multiple users report the same issue within a short period, a signal is created with the frequency count. Signals appear in the AI Panel's Signals tab and can inform roadmap decisions.

Tips for Best Results

  • Include as much context as possible in transcripts — the AI performs better with full conversations.
  • Auto-grouping works best with 10+ ungrouped entries.
  • You can always manually reassign entries after auto-grouping.
  • Use the AI Panel's Activity tab to review and approve AI-created entries.