DiPA TOOL
DiPA TOOL
Artificial Intelligence, Front-end development, Back-end development , UI design, UX design
DiPA has been designed to revolutionize the way museums approach the accessibility of digital projects. Developed for the Kainòn ETS Foundation, this tool leverages generative artificial intelligence (LLM integrated with RAG) to assess and enhance the accessibility of digital projects.
Born from a research path initiated with the project "TOWARDS a museum of the future", DiPA was built based on extensive discussions with the goal of providing the user, through a report, targeted and personalized recommendations for the specific case.
Objectives and Challenges
DiPA was created with a clear mission: to offer museums a reliable, free, and easy-to-use tool for evaluating the digital accessibility of their platforms.
The main challenge? To create a system that integrates generative AI with a simple and inclusive user interface, and to develop a secure and efficient authentication system.
Creating a Smooth and Accessible User Experience
We designed the user interface (UI) starting with interactive wireframes, used to define the structure and navigation of the site.
This process allowed us to create a smooth, intuitive, and accessible user experience, reflecting the visual identity of the Kainòn ETS Foundation and the concept 'TOWARDS a Museum of the Future.'
The interface is thus coherent, recognizable, and optimized for all devices.
Technology and Innovation: AI at the Core
The DIPA website was developed with Next.js to ensure high performance and scalability, and with Strapi CMS for a flexible and easily updatable backend. What sets it apart is the integration of a generative AI system (LLM) with Retrieval-Augmented Generation (RAG): the system compares the user's responses with a selected scientific bibliography, ensuring reports based on reliable sources and preventing hallucinations.
We defined a structured process for prompt engineering, with careful writing and testing of both system and user prompts, and the use of one-shot prompting to obtain relevant recommendations.
The final report is refined through inference in four key sequences: 1. Design, 2. Target & Stakeholders, 3. Technologies, 4. Barriers and Facilitations. Parsing strategies optimize the translation and transcription of the information provided by users.
To improve management and accessibility, DIPA includes an administrative area for users, roles, and surveys, a dashboard for data analysis, and a personalized user area for consulting and managing reports.
DiPA is not just a tool, but a strategic resource that enhances museum accessibility through AI and user-centered design.
The project is constantly evolving, with an updated RAG archive and the use of LLM always aligned with the latest available versions. We continue to actively support it, ensuring timely updates and interventions, such as the recent replacement of an API to resolve a crash.
Let’s find out if it’s a match
contact us
