We’ve added a new conversational feature to Ariadne, our LLM-enhanced guide for the Graphileon platform. It makes learning the platform more intuitive and accessible. In addition, every question that is answered, enriches the knowledge graph.
How It Works
Users can start with a straightforward question, for example “How do I send data from a query to a table?”. Then, after heaving received an initial answer, gradually build on it. The next question might be about connecting that data to a dashboard, or adding a chart. The conversation naturally evolves as complexity increases, without overwhelming the user.
Grounded in Platform Knowledge
Ariadne rephrases user questions into prompts that align with Graphileon’s core concepts, which are structured as ontologies. When the LLM generates answers, those answers are validated against the same ontologies to ensure accuracy. This means guidance is reliable and stays true to how the platform actually works.
In the background, Graphileon creates a graph that connects prompts, classifications, user comments, design patterns, function types, and answers. This structure helps organize knowledge and ensures consistency across interactions.
Why It Helps
Learning a new platform can be time-consuming. Ariadne makes it easier by providing conversational guidance that meets users where they ar, whether they’re new to data visualization or working on advanced analytics. It’s a more natural way to explore features and build skills incrementally.
We think this approach to guided learning has real potential for helping users get value from Graphileon faster.
Is this AI ?
You may have noticed that I try to avoid the term “AI”. Because I do prefer the term “Simulated Intelligence”, putting some distance between us humans and the computers and algorithms that we have granted access to LLMs and our knowledge. Computers and algorithms that constantly need to be checked and (re-)grounded, which can be achieved best using a symbolic approach. So in fact I think this is at best Symbolic Simulated Intelligence.
Docker
Cloud