South AfricaFinance and insurance2025 – ongoingNutun
Zoey, an AI voice agent for debt collection
Nutun's AI voice agent Zoey collected almost R7 million in a month on the company's own debt book
Problem
Nutun, a South African collections and business process outsourcing company, collects on large debt portfolios by phone. Human agents can only make a limited number of calls, and their call quality had to be checked.
Approach
Nutun built Zoey in-house. The team assembled existing large language models and added its own collections analytics on top. Zoey verifies a customer's identity, negotiates a payment arrangement and concludes it without a human agent. It passes complex cases to a live agent with a summary. It also predicts how likely a customer is to pay and adjusts its tone during the call. A second function scores every human agent call against more than 100 measures. Zoey has worked Nutun's R30 billion principal portfolio, more than 80,000 accounts, for about a year, in English. Nutun planned to offer it to outsourced collections clients from July.
Results
Takeaway
Nutun ran the agent on its own debt book first and used the same model to score human calls. The report gave no figure for collections before Zoey.
Categories
- Industry
- Finance and insurance
- Business function
- Customer serviceFinance, risk and legal
- Innovation type
- Process innovation
- AI technology
- AI agentsConversational and voice AI
- Adoption
- Built in-house
- Evidence
- Company's own claim
Sources
Source summaries
ITWeb reported on the launch of Nutun's AI Lab in Johannesburg and its agentic collector Zoey. It gives Nutun's figures for monthly recovery, call volume and amounts collected, and says the system was built in-house on existing language models. It also describes the call-scoring function.
Full details are in the original sources.