The challenge
A maritime fleet management software vendor ran a learning platform with more than 3,000 courses, but discovery was poor. Learners searching by keyword often couldn’t find the right course, and without specialist guidance, certification mismatches were common. Vendor staff absorbed the gap, fielding a steady stream of support queries just to point learners toward the right training.
That pattern doesn’t scale. As the course catalogue and user base grow, a support team answering the same kind of question over and over becomes the bottleneck, not the courses themselves.
The solution
NexaQuanta built an agentic AI RAG assistant that indexes all 3,000-plus courses and their metadata, giving learners a conversational interface for tailored course recommendations instead of a keyword search box. The assistant runs as a React JS plugin embedded directly into the platform, fully self-service with zero staff involvement required.
The system runs on Python and a MERN stack, using an OpenAI LLM with LangChain for the retrieval and recommendation layer, deployed as a web widget on AWS cloud.
The results
Course discovery: fully self-service across 3,000+ courses.
● Thousands of users globally now find courses through conversational recommendations rather than keyword search.
● Vendor staff support workload dropped significantly for training-related enquiries.
● Learner-to-course matching accuracy improved, boosting enrolment and completion rates.
What this means going forward
It replaced the support workflow itself, right where learners were getting stuck. Over 3,000 courses are now discoverable without a single support ticket, and the vendor’s staff can focus on work that actually needs them.
Ready to see what’s possible in your software?
Explore our Agentic SaaS Solutions →
