In this discussion with Ade Fatukasi and Selom Mortty from Sarah we explore how agentic AI could reshape credit and collections, moving the industry beyond task automation towards configurable digital employees.
The conversation examines conversational outreach, decision-making, compliance monitoring, vulnerability identification, customer transparency and human oversight.
It also considers implementation strategy, workforce concerns, AI-to-AI communication, API-enabled services and the competitive threat posed by AI-native businesses.
Find out more about Sarah -> Here.
Key Take Aways
- AI in collections is progressing from basic automation and machine learning towards conversational, decision-making and fully agentic systems.
- Industry sentiment is shifting from questioning whether to adopt AI towards determining how to implement it effectively.
- Believable conversational voice technology is expected to accelerate mainstream adoption, but also strengthens the case for transparency.
- Customers should be informed when they are interacting with AI, particularly as artificial voices become harder to distinguish from humans.
- Successful implementation requires guardrails, workforce education and careful management of employee concerns about replacement.
- Organisations should begin with controlled pilots in selected portfolio segments before committing to wider deployment.
- AI can operate flexibly as an outreach engine, compliance officer, payment manager or end-to-end collections resource.
- Quality assurance represents a significant opportunity because AI can analyse interactions systematically and identify compliance or vulnerability issues.
- Human oversight remains essential in financial services, even where multiple layers of AI monitor processes and other AI systems.
- AI-to-AI interactions may create unproductive loops, requiring workflows that detect automated responses and trigger human intervention.
- The business case is broadening from reducing cost and headcount towards increasing capacity, improving capability and enabling earlier intervention.
- Established firms face pressure to modernise as AI-native competitors build new operating models without legacy systems, processes or technical debt.
Innovation
- Positioning AI as a configurable digital employee rather than a conventional software tool.
- Using AI to determine the most effective contact method and timing based on customer and account information.
- Applying conversational analysis to evaluate compliance, identify vulnerability and improve customer treatment.
- Monitoring more than 100 data points within a telephone interaction.
- Designing layered assurance models in which AI monitors other AI, while a human retains final decision-making authority.
- Detecting AI-generated customer communications and automatically routing cases for human intervention.
- Enabling natural-language instructions to initiate complete collections workflows.
- Using APIs to connect AI with payments, account reconciliation and other business services.
- Configuring collections agents around an organisation’s own tone, rules and customer language.
- Using external data, including Companies House information, to support commercial targeting and business development.
- Employing AI-assisted coding and low-code tools to reduce the resources needed to build and operate a technology company.
- Developing AI employees that understand the full context of a business, including historical cases, previous problems and future plans.
Key Statistics
- The speaker began exploring the AI space six or seven years ago.
- An AI system can assess more than 100 data points during a telephone call.
- Some organisations want technology capable of servicing hundreds of thousands of cases simultaneously.
- The speakers estimated that fully natural-language execution of tasks may be four or five years away.
- Sarah’s technical co-founder reportedly did not expect to hire another engineer over the following two quarters.
- The speakers said the company could not have been started in the same way 18 months earlier.
- Three people were said to be operating a company that might have required ten people six years earlier.
- The discussion considered how AI adoption could develop over the next five years.
- A hypothetical natural-language payment instruction involved sending £25 to another account.
Key Discussion Points
- The evolution of collections technology from rules-based automation to agentic AI.
- The distinction between automating existing tasks and redesigning entire collections processes.
- How AI can learn when, how and through which channel to contact customers.
- The importance of disclosing AI use during customer interactions.
- Changing customer and corporate attitudes towards conversational AI.
- Safe implementation through guardrails, education and incremental deployment.
- The role of AI as both an employee-support tool and an autonomous operational resource.
- AI-enabled quality management, compliance monitoring and vulnerability identification.
- The continued requirement for human accountability in sensitive financial-services decisions.
- The growth of API-based services and machine-to-machine business activity.
- The risk of AI-generated complaints creating automated response loops.
- Competitive pressure from AI-native firms and the need for established collections businesses to experiment now.
#Sarah
RO-AR.com contact list
