HOW SANIA WORKS

From a visitor question to the avatar’s answer

SANIA brings several components together in a simple visitor experience: interact, understand the request, retrieve relevant information, optionally use a tool, then deliver an answer through the avatar.

A simple front-end experience, several layers behind it

For visitors, the goal is natural interaction: ask a question and receive an answer. The technical complexity stays behind the interface.

Depending on the project, SANIA combines voice or touch interaction, session context, an LLM, a knowledge base and configured tools.

A simplified question journey

01

Interaction

The visitor speaks or uses the interface.

02

Understanding

The request is interpreted in conversation context.

03

LLM

The conversational engine determines how to handle the request.

04

RAG / knowledge

Relevant business information can be retrieved.

05

Tools

A function, API or webhook can be used if required.

06

Answer

The result is turned into a conversational response.

07

Avatar + voice

SANIA delivers the answer to the visitor.

Four layers to remember

01

Interface

Display, kiosk, web or application interface depending on deployment.

02

Conversation

Voice, session context, interruptions and languages.

03

Intelligence

LLM, instructions, knowledge base and document retrieval.

04

Tools

Functions, APIs, webhooks and specific actions when included in the project.

Architecture depends on the project

SANIA can use different LLMs and tools depending on the configuration. One architecture is not imposed on every use case.

A platform capability should not be confused with an integration that already exists. APIs can be used, but a specific CRM is only connected when that integration has been implemented.

FREQUENTLY ASKED QUESTIONS

Frequently asked questions

Does SANIA need a predefined dialogue tree?

No. SANIA uses conversational AI and can understand naturally phrased questions, while the permitted information and action scope should still be clearly defined.

How does SANIA find information in a document?

SANIA can use semantic retrieval / RAG to find relevant information in knowledge-base content before building an answer.

Can SANIA use different LLMs?

Yes. Different models can be used depending on project configuration and architecture.

SANIA

The best explanation is still a conversation

Try SANIA directly and see how the visitor experience remains simple.

See how this architecture can fit your project

Tell us about your context, knowledge and potential integrations. We can prepare an appropriate demonstration.

Request a demo