An AI knowledge base that gives SANIA the context of your organisation

To welcome and inform visitors correctly, AI needs information that reflects the location where it is used. General knowledge is not enough when a question concerns service hours, reception points, available services, procedures, products or rules specific to an organisation.

The knowledge base defines this context. Your organisation chooses what should be accessible and provides SANIA with the relevant content. The avatar can then build answers within that scope rather than behaving like a general-purpose AI disconnected from the real venue.

This is particularly important in physical spaces, where visitors expect practical answers they can use immediately. Clear and current content helps keep the experience aligned with the reality of the organisation.

A knowledge base is not a script for every possible question. SANIA uses conversational AI, so visitors can express requests naturally and the avatar can answer using the context and information made available to it.

This avoids relying entirely on rigid menus or dialogue trees. The organisation does not need to predict every exact sentence a visitor may use. The priority is to define useful information, its scope and its quality, then learn from the questions encountered in real use.

Different LLM engines can be used depending on the project configuration. However, the model itself cannot replace organisation-specific knowledge: even a capable model still needs accurate context about your location and services.

More content does not automatically mean better answers. A useful knowledge base focuses on information relevant to the role of the avatar and maintains a clear boundary between what SANIA can explain, what should remain with staff and what may require an external system.

Editorial maintenance therefore matters. When opening hours, services, offers or procedures change, the relevant content should be updated. Each organisation can define its own validation process according to teams and information sensitivity.

Representative questions can also be tested before deployment and after major content changes. The goal is not to write every conversation in advance, but to verify that essential information is present and understandable.

Knowledge supplied to SANIA is different from a live integration with a business application. Adding documents does not mean the avatar automatically queries a CRM, ERP, ticketing platform, booking system or internal software.

If a project requires data from a third-party system or an action inside that system, the connection must be assessed separately. A specific integration may be considered depending on the need, but it should not be confused with the native knowledge-base function.

Keeping this distinction clear makes the project more credible: the knowledge base makes organisation-provided information accessible, while external-system exchanges are a separate integration requirement.