AI AT THE HEART OF KNOWLEDGE MANAGEMENT AND PROFESSIONAL TRAINING
— A chatbot and condensed videos to speed up the sharing of skills between co-workers thanks to artificial intelligence.
At a time where companies are having to manage a growing mass of content for in-house training, Skillagora is tackling a major challenge: structuring and adding value to this information to make it easier for learners to adopt and use.
To meet this challenge, BeTomorrow has developed an artificial intelligence capable of transforming these data streams into relevant, usable content. While automatic transcripts are often cluttered with superfluous elements that make them difficult to use, BeTomorrow's AI comes into play to effectively synthesise this training content.
Users can upload a video to the Skillagora space and the AI will extract the essential information, making it easier to learn skills and share knowledge.
The AI solution developed by BeTomorrow aims to:
Facilitate the absorption, consultation and use of training content between peers, by helping employees to quickly understand the content of the videos.
Provide learners with quick and convenient access to information and training by offering an instant summary of these video resources.
Enable the indexing of video content, which is basically mainly oral, to make it easily accessible via the search bar.
Improve employees' discovery of relevant training content by using advanced search and recommendation technologies.
Automatic video summary
Our solution incorporates a powerful LLM (Claude) model to generate a precise summary of videos in a matter of seconds as soon as they are downloaded. This feature allows employees to save precious time by accessing the essential points of the content directly. Each summary can be easily modified or re-generated to refine the output and ensure that the summary is perfectly adapted to the user's needs.
Automatic video summary
Our solution incorporates a powerful LLM (Claude) model to generate a precise summary of videos in a matter of seconds as soon as they are downloaded. This feature allows employees to save precious time by accessing the essential points of the content directly. Each summary can be easily modified or re-generated to refine the output and ensure that the summary is perfectly adapted to the user's needs.
Advanced semantic searching
Our semantic search engine uses query vectoring to go beyond simple keyword matches and identify related topics, even if the exact terms searched for are missing. This approach guarantees more relevant and contextual results, optimising access to shared information and skills.
Advanced semantic searching
Context-sensitive recommendation system
Our contextual recommendation technology automatically suggests related subjects based on the semantic space of the platform. By displaying relevant and complementary content, it encourages exploration and enhances the user experience by making it easier to discover new and relevant resources.
Context-sensitive recommendation system
Conversational approach (Skillbot and RAG)
The integration of a chatbot based on an artificial intelligence RAG (Retrieval-Augmented Generation) model enables the knowledge repository to be queried in natural language. Learners can ask questions and obtain contextualised answers, providing immediate and relevant access to available content.
Conversational approach (Skillbot and RAG)
Encapsulating the user experience
Our solution provides complete automation in the background, guaranteeing a fluid and intuitive user experience. Interaction with the training repository is seamless, requiring no technical skills, giving everyone easy access to the resources they need.
Encapsulating the user experience
A completely new learning experience: The artificial intelligence solution developed and deployed by BeTomorrow on Skillagora transforms the peer-to-peer learning experience by drastically reducing the time required to assimilate content.
Considerable time savings for employees: This internal, business-specific, high value-added training is often ‘oral’. Using advanced automatic summarisation algorithms, it condenses 1.5 hour training videos into just 30 seconds. This optimisation encourages greater adherence and retention on the platform, while guaranteeing that the essential information is faithfully reproduced.
Optimum scalability: our generative AI ensures smooth deployment in production and has proven reliability. Based on refined search and recommendation models, the solution continually improves the relevance of the content offered. The integration of conversational elements enriches interaction with learning resources, making learning more dynamic and engaging.
Rigorous performance evaluation: Using precise indicators - in particular the quality of summaries and recommendations - our approach guarantees measurable results and an optimised user experience
Rate seen by AI-based personalised learning platforms compared to traditional learning methods (Source: McKinsey, 2022).
Companies using AI-powered training tools reduce their need for physical media, trainers and resources (Source: Training Industry, 2023).
AI-based platforms enable employees to access relevant learning content quickly and efficiently (Source: Forrester, 2021).
How can AI reinvent training and knowledge sharing?
"This project is a concrete example of how AI can address a very real business need: making in-house training more accessible, faster and more engaging. By combining technologies such as automatic summarisation, semantic search and RAG, we have designed a fluid, user-centred experience that is free of technical complexity. Skillagora doesn't just automate: it structures knowledge and facilitates its transmission between peers. It's a robust piece of technology, but designed for human use. And that's where AI comes into its own."
Gilles Mantel
Product Manager
BOOST YOUR AI STRATEGY WITH OUR GENAI WORKSHOPS
- A clear and accessible approach to understanding how AI works and its potential.
- An innovative card game to raise awareness and train your teams to generate relevant and concrete AI use cases.
- An evaluation methodology to assess the feasibility, complexity, and impact of your AI projects.
- A personalized debrief with an actionable, prioritized roadmap for a fast and effective implementation.
How does BeTomorrow's AI optimise the use of transcripts?
The AI developed by BeTomorrow takes over the processing of transcripts by sending them to a language model (LLM), which produces a clear, structured summary. The process is based on a specific prompt, asking the LLM to generate a concise 10-line summary based on the context of the content. To ensure consistent results and avoid model "hallucinations", BeTomorrow has refined the prompt system to produce accurate and consistent summaries.
How can AI improve training and e-learning?
What are the key benefits of integrating AI into training solutions?
