AI recommendation system Lumen
Conversational recommendation system
With an AI assistant, its own knowledge base, custom models and multi-level recommendations.
The project
We built it for an online platform where users don’t choose from categories but say what they need, and the system recommends based on the conversation. The system has to understand, all at once, the user’s needs, the trained knowledge base and the items it can recommend – and the method can be trained for any field.
What we built
01Conversational AI assistant: users describe what they’re looking for in their own words, without categories or filters
02Own knowledge base: the AI works from a knowledge base we built and trained for the field
03Custom models: models built specifically for the task
04Multi-level recommendations: based on the conversation and the user’s needs, the system makes a specific suggestion
05Recommendation database: the set of items the system draws its suggestions from
06Secure operation: can be deployed on your own infrastructure or a private cloud, with its own model, so internal data never leaves
Domain expertise gained
Secure conversational AI built on its own knowledge base, from training to specific recommendations.
Conversational recommendation (CRS) RAG architecture Building a proprietary knowledge base Custom models Own model in a closed environment Multi-level recommendation system GDPR-compliant data handling Cybersecurity audit and hardening
Where it can be used
IN THIS PROJECT Online platform
Conversation: user → AI assistant Result: the best-fitting recommendation from the offering
Enterprise internal knowledge base
Conversation: employee → internal policies, documents Result: accurate answers from the company’s own knowledge, in a closed system
Customer service
Conversation: customer → product and contract data Result: a solution or the right product
Online shop
Conversation: shopper → product catalogue Result: the right product
Financial services
Conversation: customer → products and terms Result: the right loan or insurance
HR and onboarding
Conversation: new employee → internal processes Result: the next step and the colleague responsible


