Each preservation record includes an ID, method name, description of the technique, and the safe temperature range. For example, refrigeration returns '0 degrees C to 4 degrees C' with a description explaining how low temperatures slow bacterial growth. The agent formats this data into conversational, easy-to-follow guidance.

The agent retrieves all available methods from the API and uses its training context to match the right technique to the food item mentioned. For example, if a customer asks about preserving berries, the agent filters results to suggest freezing, dehydration, or canning based on suitability and available temperature ranges.

Access control depends on your agent configuration. You can restrict write operations (POST, PUT, DELETE) to authenticated internal users, such as food safety officers or quality managers, while keeping read operations (GET) open to all customers. Set role-based permissions in your Tars gambit settings.

The list endpoint returns all preservation methods with their temperature ranges. Your AI agent can filter these results conversationally. When a customer says 'I only have a room temperature kitchen,' the agent identifies methods like drying, salting, or canning that do not require refrigeration from the returned data.

All API calls use your X-API-Key for authentication. Data is transmitted over HTTPS and queried in real-time during conversations. Tars does not maintain a separate copy of your preservation database. The information is fetched live and presented only within the active conversation context.

The API stores preservation methods with general temperature ranges and descriptions. If your database includes region-specific entries or notes, the agent can present that information. For regulatory compliance questions beyond what the database contains, the agent can escalate to a food safety specialist.

The agent first searches the API for matching records. If no results are found, it can use its general training knowledge to provide a preliminary answer while noting that the method is not in the verified database. You can also configure it to suggest adding the method through an internal review process.

A static FAQ is fixed and one-directional. This integration lets customers ask follow-up questions, compare methods, and get personalized recommendations based on their situation. The agent queries live data, so when preservation records are updated, every future conversation reflects the latest guidelines instantly.