With the recent launch of OpenAI’s “omni” model, we are now deep in the era of Generative AI. The capabilities of artificial intelligence have been greatly expanded, with large language models performing a variety of complex tasks such as code writing, art creation, quality assurance, and even fraud detection. As society adapts to these changes, the collections industry must also embrace Generative AI and explore its vast potential to solve common challenges.

Technological booms have always been met with cautious skepticism. From the introduction of steam engines to the invention of the telephone, society has often welcomed innovation with a degree of reservation. However, if history teaches us anything, it is that early adopters, innovators, and pioneers also approach technological innovations with courage. Artificial intelligence is here to stay—and it’s here to change the world. The question is, which problems are we trying to solve, why, how, and when?

In the accounts receivables industry, we have seen the emergence of several early adopters among creditors, lenders, and collection agencies. AI tools are already in use to monitor compliance, automate consumer conversations, analyze credit risks, and build financial models. With time, we are seeing the ROI of these tools increase. How should collection leaders approach this critical phase?

Achieving Collection Performance Beyond Human Capabilities

Banks and fintech companies are progressively investing in technology to gain a competitive edge. Augmenting manpower with AI has improved staff efficiency and productivity while driving down costs; concurrently, better consumer engagement has led to topline growth.

In the Conversational AI space, we are automating consumer conversations across all channels—voice, text, chat, and email. Commonplace tasks such as consumer verification, payment negotiation, and payment processing can now be automated in both spoken and written interactions, enabling live agents to focus on complex queries requiring additional research, expertise, and skills. This change is helping creditors and collection agencies improve their collection rates and agent productivity and reduce the cost of collections.

This is not aspirational thinking; what I’m describing is taking place right now. My prediction is that Conversational AI will automate 90% of consumer interactions within the next two years.

Why LLMs Represent a Transformative Solution for Financial Services Organizations

Large language models are fed with immensely large amounts of public and proprietary data and continue learning from ongoing interactions and new data inputs.

Collection agencies' customer relationship management tools (CRMs) contain roughly two decades of consumer interactions, consumer profiles, transactions, and agent notes. This data goldmine can help us better understand the consumer journey, behavior, and payment propensity.

Today, financial services organizations rely on credit scores to evaluate consumers’ creditworthiness. However, credit scores have two main blind spots:

  1. The evaluation models are based on limited credit lending history and are therefore prone to bias against specific consumer segments with limited credit histories.
  2. The models do not contain granular data on consumer behavior for effective engagement (e.g. the best time to contact the consumer).

Large language models, when trained with consumer data and past agent interactions, can bridge these gaps and curb biases, effectively helping financial services organizations determine the best engagement and recovery strategy at an individual level. We can refer to these LLMs as large collection models.

In the future, orchestration platforms for the debt collection industry—collection orchestration platforms—will become instrumental when trained with consumer engagement data from creditors’, lenders’, and collection agencies’ CRMs. The data will help the AI tools evaluate the consumers’ propensity to pay, accelerating the recovery process and enabling collectors to allocate more resources for the most complex segments. This strategy can help optimize processes, maximize recovery rates, and cut collection costs.

Here are a few ways engagement strategy and consumer experience (CX) can be optimized based on granular data and past engagements:

  • Identify the consumer’s preferred communication channel (text, email, voicebot call, live agent call, print letter, etc.)
  • Analyze the best day and time for engagement
  • Determine the resources required for account resolution

Collection orchestration platforms are going to become more effective with time, as more consumer engagement data generated by their use enables them to self-train further, automating the channel, timing, and strategy of the recovery effort.

How To Leverage New Tech Advancements Effectively and Responsibly

As with any new technology, the adoption of Generative AI into the debt collection industry requires a collective effort and commitment to ethical standards and responsibility, especially with the utilization of large amounts of consumer data.

Here are a few things to keep in mind to ensure a successful implementation:

Compliance guardrails: It’s crucial to prevent hallucinations, recognize and eliminate biases, and ensure the technology is programmed to adhere to all applicable industry regulations, including data privacy measures.

Data accuracy: Generative AI is only as good as the data it’s fed. Therefore, it’s important to monitor data accuracy and live agent logs.

Seamless system integrations: To obtain the best results, seamless integration between the CRMs and orchestration platforms is required. Offline integrations and data transfers pose a risk of data loss and privacy breaches.

In Conclusion: From “if” and “why” to “when” and “how”

Collection orchestration platforms are poised to shape the future of the collections industry. Industry leaders should focus on these technological advancements and look beyond the capabilities of current large language models. Today, the adoption of Generative AI is no longer a matter of “if” or “why” but rather “when" and “how.” The rapid pace of new developments in LLMs necessitates this forward-thinking approach.

While today the focus is on LLMs, the future of collections will center around large collection models and collection orchestration platforms. It’s time for the accounts receivables industry to begin a collective effort to integrate these advanced technologies and redefine best practices.

To learn more- Speak With Our Conversational AI Expert.

Next Article: Opting Out of Letters? Not So Fast ...