Solution type

Digital Assistants

Conversational digital agents that answer questions, provide guidance or lead users through a process in text, voice or as a 3D character. The focus here is the interaction between the user and the assistant.

The problem it solves

On websites, at reception desks, in showrooms and at exhibitions, the same questions appear repeatedly while a human specialist may not always be available. A digital assistant can provide natural-language guidance, but it is only credible when it works from clearly defined knowledge and can signal uncertainty. If the main requirement is organising and searching large document collections, an AI knowledge system is the better starting point.

When it makes sense

Reducing repetitive work at reception or information points

Providing a product specialist at exhibitions

Answering internal questions in natural language

Creating an instructor or simulation character

Building a museum guide or historical character

Offering multilingual visitor or customer guidance

Key benefits

Natural-language access to complex information

Consistent answers across locations and interfaces

Text, voice and 3D character formats can be combined

Role and communication style can match the use case

Repetitive questions can require less staff time

Critical cases can be routed to a human handover

Limitations and design considerations

Assistant quality depends directly on the accuracy and maintenance of its knowledge base. Users should be clearly told that they are interacting with an AI system, and the design must cover questions for which no reliable answer is available. Healthcare, safety and other critical topics require human escalation. Conversation storage, access and personal-data handling also need explicit rules.

Related technologies

AIAI charactersVoice TechnologyTouchscreenVR

Typical features

Text conversation

Speech recognition and playback

3D character and animation

Multilingual operation

Controlled knowledge base

Uncertainty handling

Human handover

Conversation logs and analytics

Project examples

ARWorks · 2023

BOB - the all-knowing virtual assistant

An AI-powered virtual assistant and conversational partner that can be given its own personality

Open →

ARWorks · 2024

ChatGPT-based 3D virtual colleagues

Hostess, receptionist, salesperson, product expert and others

Open →

Medtronic · 2025

Medtronic — McGRATH VR Training

A Quest 3 VR simulator with AI-powered patients teaching the use of the McGRATH videolaryngoscope.

Open →

Related ARworks platforms

TouchAIAI-enabled multimedia knowledge and customer-service platform for touchscreen environments.Services →

Q&A

What information can a digital assistant answer from?

The system can work from approved company documents, product data, manuals or structured data sources. Reliability improves when the organisation defines which sources are authoritative and how they are kept up to date. Access to the open web is a separate design decision because it introduces different verification, privacy and control requirements.

Can users speak to the assistant?

Yes. Speech recognition converts the user's voice into text, the AI prepares a response and text-to-speech can read it aloud. Voice use needs attention to background noise, response latency and personal-data handling. At exhibitions or other noisy venues, a text-based fallback is often useful even when voice is the primary interaction.

What happens when the AI is not sure of an answer?

A well-designed assistant should not respond confidently to every question. It can ask for clarification, state that it does not have enough information, show the relevant source or route the user to a human colleague. For critical workflows, this behaviour should be part of the original specification rather than an add-on introduced after launch.