How Santander optimized its portfolio management with conversational AI

Banco Santander managed early-stage credit card delinquencies through external agencies that operated with call centers and human agents. This model presented structural challenges for a large-scale operation:

• High operational costs associated with human management.
• Limited scalability: volume growth required more agents.
• Slow processes and low personalization in customer contact.

The bank needed an alternative that could match or surpass human performance, while maintaining operational efficiency and full strategic control.

Colektia's AI infrastructure was implemented, integrating its AI agent as a 100% automated collection platform, without human involvement.

The strategy relied on:

• Digital omnichannel model: SMS, email, IVR, WhatsApp, and AI voice agent.
• Automatic personalization: AI defines the optimal channel, time, and frequency for each customer.
• Continuous learning: each interaction feeds the model to improve future effectiveness.

To evaluate actual performance against the traditional model, a comparative pilot was designed for 3 months (August – October 2025).

Pilot structure:

• 12,000 early-stage credit card delinquency accounts.
• 6,000 accounts managed by Colly, 100% with AI.
• 6,000 accounts managed by traditional agencies, with external human agents.

Both groups operated under the same delinquency window and comparable conditions.