United StatesManufacturing2025 – ongoingLippert
RAG assistant and call scoring for customer support
Lippert cut new support agent ramp-up from six months to four weeks with a generative AI assistant
Problem
Lippert, a manufacturer that supplies components for recreational vehicles, handles more than 1 million support calls a year. Managers reviewed only about 100 calls a month by hand, and new agents needed months to learn the products.
Approach
Lippert built a retrieval-augmented chatbot on Databricks. It draws on product manuals, case histories and field expert videos and guides agents during calls. The company also scores calls with AI, transcribes and summarizes them, and sends the summaries to Salesforce. Models were tested and monitored with Databricks Agent Bricks, using synthetic test data.
Results
Takeaway
AI call scoring let the company review all calls instead of a small sample. The assistant was also used as a training aid for new agents.
Categories
- Size
- Large enterprise
- Industry
- Manufacturing
- Business function
- Customer service
- Innovation type
- Process innovation
- AI technology
- Generative AI
- Adoption
- Off-the-shelf tool or platform
- Evidence
- Vendor case study
Sources
Source summaries
Databricks describes Lippert's retrieval-augmented support assistant and its AI call scoring and summarization. It reports an 85% cut in new agent ramp time, from six months to four weeks, and model accuracy rising from 33% to 84%. It says the company now reviews all of its more than 1 million yearly calls instead of about 100 a month.
Full details are in the original sources.
