Periodic Reporting for period 1 - ASSIST (Aware Systems for Smart Interactive muSeums and arT)
Okres sprawozdawczy: 2025-02-01 do 2025-06-30
The ASSIST project, funded under the EIC Booster scheme, aimed to accelerate the exploitation of advanced AI technologies originally developed within the EIC Pathfinder project ASTOUND. ASSIST focused on adapting these innovations for the cultural tourism sector, with museums as a primary use case. The project addressed growing demand for personalised, accessible, and immersive visitor experiences by developing an intelligent chatbot capable of interacting in real time, in multiple languages, and across various visitor profiles. Cultural tourism is a rapidly expanding global market, with Europe at its forefront. Museums in particular are increasingly investing in digital transformation, yet many lack dynamic tools that can adapt to diverse audiences. ASSIST was designed to fill this gap by creating a context-aware, multimodal conversational agent that enhances learning and engagement while respecting curatorial integrity. The overarching objectives of ASSIST were to demonstrate the feasibility of this technology, validate its market potential, and prepare for future commercialisation. The project sought to deliver a minimum viable product (MVP), supported by stakeholder insights, a structured dataset, and a robust technical prototype. Ultimately, ASSIST aimed to set the foundation for a scalable, ethical, and high-impact AI solution that can transform how cultural heritage is experienced and communicated.
Over its five-month duration, ASSIST successfully developed a functional, multilingual chatbot Demo capable of voice, text, and image-based interactions. This system integrates personalised dialogue, image recognition, and content recommendations, adapting its tone, depth, and language to different visitor types, such as children, tourists, or experts.
Key activities included:
- A market analysis of cultural tourism, identifying a €2.3 billion addressable market, with museums as a high-potential entry point.
- Stakeholder engagement involving 35 museum professionals across Europe, which helped shape the MVP and its feature set.
- The creation of a structured, museum-specific dataset stored in a graph database, ensuring content consistency and adaptability.
- Generation of over 6,000 synthetic dialogues using personalised prompts, tested through human and AI-based evaluation methods.
- Development of a Demo of a modular chatbot architecture using hybrid language models, multimodal input/output, and an AI “shield” to monitor quality and safety.
- A public demonstration during a focus group with 30 cultural leaders confirmed the system’s potential and generated strong interest in future pilots.
These achievements demonstrate ASSIST’s technical viability and commercial promise, providing a solid base for further development and uptake.
Key activities included:
- A market analysis of cultural tourism, identifying a €2.3 billion addressable market, with museums as a high-potential entry point.
- Stakeholder engagement involving 35 museum professionals across Europe, which helped shape the MVP and its feature set.
- The creation of a structured, museum-specific dataset stored in a graph database, ensuring content consistency and adaptability.
- Generation of over 6,000 synthetic dialogues using personalised prompts, tested through human and AI-based evaluation methods.
- Development of a Demo of a modular chatbot architecture using hybrid language models, multimodal input/output, and an AI “shield” to monitor quality and safety.
- A public demonstration during a focus group with 30 cultural leaders confirmed the system’s potential and generated strong interest in future pilots.
These achievements demonstrate ASSIST’s technical viability and commercial promise, providing a solid base for further development and uptake.
ASSIST advances the state of the art by combining several innovative elements into a single, deployable system: i) A context-aware AI engine that adapts responses based on visitor profile, location, and behaviour; ii) Integration of multimodal input (speech, text, image) and retrieval-augmented generation for rich, dynamic interactions; iii) Tools for curatorial oversight and safety modules to manage AI risks and prevent hallucinations or inappropriate content. Unlike existing digital guides, ASSIST enables museums to offer real-time, adaptive experiences that can be curated, personalised, and scaled. Its modular design supports flexible deployment models—on-premise or cloud-based—ensuring compatibility with diverse institutional needs. To ensure further uptake, ASSIST will require continued investment in: i) Large-scale piloting in real museum settings to validate usability and impact, ii) Robust multilingual support, including co-official European languages, iii) Commercial partnerships to refine the business model and reach new market segments, iv) compliance with evolving AI regulation and data protection standards. The project lays the groundwork for future commercialisation through the EIC Transition programme and opens pathways to adoption in adjacent sectors, such as archaeological parks and heritage tourism.