The three levers
| Lever | When to pull | What to change |
|---|---|---|
| Give knowledge | The AI doesn’t know the answer or gives outdated/wrong information | Add or fix Snippets, Data Providers, or Search Tables |
| Instruct behavior | The AI has the right information but uses it poorly (wrong tone, skips steps, doesn’t escalate) | Add or refine Guidance rules or Procedures |
| Add system access | The AI needs to read or write data in an external system | Connect a Toolbox, MCP Server, or Unitool |
Identifying potential
Before you fix anything, find out where the agent struggles. Three approaches work well together.Suggestions
Suggestions automatically analyze your conversations and surface questions the AI failed to answer. The system clusters similar questions, matches them against your existing knowledge sources, and categorizes each cluster by root cause:| Issue type | Meaning | Lever |
|---|---|---|
| Missing content | Your knowledge base lacks the answer | Give knowledge |
| Can’t read customer data | The AI needs data from an external system | Add system access |
| Can’t write customer data | The AI needs to perform an action in an external system | Add system access |
Topics and metrics
Open Topics and look at the treemap. Large red boxes represent high-volume, low-resolution topics. Fix these first for the biggest impact. Click a topic to review its conversations and categorize the root cause. Track your Resolution Rate, Involvement Rate, and CSAT weekly. High resolution + low CSAT usually means accuracy problems (wrong answers). Low resolution + many “No Answer” responses points to knowledge gaps.
Conversation review
Filter Conversations to 1-2 star ratings or “Not Involved” status. Read 10-20 threads and sort issues into the three levers above. A lightweight weekly routine where each team member reviews a handful of conversations catches problems that automated analysis misses.Give knowledge
When the AI says “I don’t have information about that” or gives an outdated answer, it needs better knowledge. Wrong answers. Find them by filtering for low ratings with feedback like “wrong” or “incorrect.” Click the AI message and open the Improve Answer sidebar to see which sources the AI used. Fix the root cause: update the outdated document, remove conflicting sources, or clarify the passage in Data Providers or Snippets. Missing answers. The fastest fix is a Snippet: click the AI message, select Add Snippet, and write the ideal answer. For recurring topic clusters, add a full documentation source (webpage crawl, PDF, or Search Table). Suggestions with issue type “Missing content” point you directly to the biggest knowledge gaps. Before product launches or seasonal events, add documentation proactively.Instruct behavior
When the AI has the right information but responds with the wrong tone, skips steps, or answers when it should escalate, it needs better instructions. Guidance rules tell the AI how to behave in specific situations. Use audience conditions to scope rules to the right conversations and keep instructions concrete: “Always include a tracking link when answering shipping questions” works better than “Be helpful.” Procedures guide the AI through multi-step workflows. Use them when the conversation must follow a specific sequence, such as collecting order details before processing a return.Add system access
When the AI needs to look up customer-specific data or perform actions in external systems, it needs system access. Suggestions with issue types “Can’t read customer data” or “Can’t write customer data” highlight these gaps automatically. See Actions for the full overview. Reading data. The AI might need order status from your database, account details from your CRM, or inventory levels from your e-commerce platform. Choose your integration based on availability:- Pre-built Toolboxes or MCP Servers if botBrains offers one for your system.
- Unitools to write custom code that queries your API or database.
- Search Tables if the data rarely changes and you can upload it.