The agent matches the customer's question to content items using codenames, external IDs, or keyword-based listing with pagination. You can configure content mapping rules that associate common questions with specific content item identifiers. The agent can also browse content listings to find the most relevant item dynamically.

Yes. The agent detects the customer's language from the conversation context, then uses Kontent.ai's Get Language Variant endpoint to fetch the matching localized version of the content item. If a variant is not available in that language, the agent falls back to the default language and informs the customer.

The agent can work with all structured content elements that Kontent.ai returns through its API, including rich text, plain text, numbers, date-time values, multiple choice, linked items, and asset references. For images and files, the agent provides URLs. Rich text is converted to conversational language for natural delivery.

No. The agent queries Kontent.ai's API in real time for every conversation. This means customers always receive the latest published version of your content. When you update an article or product description in Kontent.ai, the agent immediately serves the new version without any cache invalidation or redeployment.

The integration primarily uses the Management API (via the Subscription API key) for retrieving content items, content types, languages, and language variants. The List Content Items endpoint uses Delivery API patterns with continuation tokens. Both read paths are supported for comprehensive content access.

Website rendering shows content on a page that visitors must find and navigate to. With Tars, the AI agent proactively retrieves the relevant content item based on the customer's question and delivers it within the conversation. Customers do not need to search your site or browse help centers. The right content finds them.

The Subscription API key provides access to list projects and environments, retrieve content items and their variants, browse content types, and manage language configurations. It is a management-level key. For production use, ensure you generate a key scoped to the specific environments the agent should access.

Yes. The List Content Types endpoint supports pagination through continuation tokens, so the agent can efficiently browse large content models. In practice, you configure the agent to target specific content types relevant to customer conversations, such as FAQ, product, or article types, rather than iterating through the entire taxonomy.