Unified AI Controls & Agent Flexibility
✨ What's new?
- New "Answer Depth" settings for Knowledge Base: We’ve introduced a new "Answer Depth" setting to give you full authority over how much information the AI reviews before generating a response. This helps you balance speed and detail based on your content structure.
- Shallow: Checks the top 4 most relevant sections. Best for fast, precise answers.
- Balanced (Recommended): Reviews the top 6 sections. The best mix of speed and accuracy for most cases.
- Deep: Examines the top 10 sections. Ideal for complex questions or long procedures.
- Select your AI Agent model: We’ve added the ability to choose which AI model powers your agent. You can now select between gpt-4o-mini, gpt-5-mini, or our recommended gpt-4.1-mini, giving you more control over speed and answer quality.
- Adjust AI Verbosity for GPT-5: When using the gpt-5-mini model, a new "Verbosity level" dropdown lets you choose the level of detail in responses:
- Low (concise)
- Medium (balanced)
- High (detailed and explanatory).
- Allow agents to delete internal notes after submission: To help agents rectify mistakes, we’ve added the option to delete internal notes once they’ve been submitted (restricted to the note owner). This ensures the case history remains accurate and error-free.
⚡️ Improvements
- Clearer unassignment messages in Chat: We’ve replaced generic system captions with descriptive reasons. When a case is unassigned due to an agent going offline, the system now clearly displays "Agent went offline," providing immediate context for team supervisors.