automatic URL capture, linkage to the knowledge base, and dynamic responses based on the active page

Business Problem

Currently, the Trengo AI Agent lacks contextual information about the specific page a customer is on. While human agents in the Trengo inbox can immediately see the URL a chat was started from, the AI agent does not have this information in its 'working memory'.

This leads to the following bottlenecks:

• Friction in the customer journey: Customers expect the agent to know what they are looking at. Questions like "How much does this product cost?" or "Is this in stock?" now result in generic answers or a request to specify which product the customer means.

• Lower conversion: The AI cannot make a personalized offer or answer specific product questions without the customer having to repeat information manually.

• Unnecessary escalations: Because the AI lacks context, simple questions are still escalated to human agents, which reduces the degree of automation.

Wished Result

We want the AI agent to have access to the 'Current Page URL' (and ideally other metadata such as page titles or custom tags) as part of the prompt context.

The ideal solution includes:

1. Context input: The URL of the page where the chat is started must be automatically passed to the AI agent as a variable.

2. Knowledge linkage: The AI must be able to match this URL with information in the connected knowledge base (e.g., a specific page in the help center or a product feed).

3. Dynamic Response: The AI must be able to respond: "I see you're looking at the [Product Name]; its price is €X, and we have 3 more in stock."