Brandwise gives you several ways to improve AI reply accuracy over time. The more feedback and context you provide, the better the AI gets.
Two Types of Feedback
General Feedback
Broad rules that affect how AI drafts replies across all conversations
Used to define tone, language preferences, and consistent behaviors
Interaction Feedback
Case-by-case corrections on individual AI replies
Used to fine-tune the AI for specific scenarios
Adding General Feedback
Go to AI Assistant > Feedback to add broad instructions. Examples:
"Do not use emojis in any replies."
"Keep responses professional and friendly."
"Always greet the customer by name if provided."
Adding Interaction Feedback
After generating an AI reply, a popup appears where you can enter feedback on that specific interaction. Use this when:
The AI reply was partially correct but needs improvement
The AI repeatedly handles a certain type of comment incorrectly
You want to adjust how the AI handles this scenario in the future
Response Examples
The AI learns from the replies you send. When you generate a reply, edit it, and send it, that reply is stored as a Response Example. Over time, the AI uses these examples to match your tone and style.
You can view, edit, and delete response examples in your brand settings. Remove outdated examples if your messaging has changed.
Leveraging Product URLs
For ad comments, the AI automatically knows which ad and landing page the comment is from. To help it answer product-specific questions accurately, include your product URLs in your Knowledge Base using a format like:
Product Title: [title]
Product URL: [link from ad]
Product Description: [info, pricing, etc.]
This helps the AI answer questions like "How much is this?" with accurate product-specific information.