Digital and AIPreparing for AI in collections and recoveries
Artificial Intelligence is rapidly moving from concept to practical application across financial services.
Forid Meah · 16 March 2026
AI
AI strategy, governance and implementation
Your board wants an AI plan and your regulator wants to know who owns every outcome the AI influences. We help you answer both, and take AI into live operation safely.
47
AI providers evaluated worldwide, so your shortlist starts from the whole market.
Source: arum global AI evaluation programmeWhere AI succeeds
AI success starts long before implementation.
AI in collections means using models to prioritise contact, identify vulnerability and support decisions, with governance that can explain each one.
AI works in collections when three things are in place first: clear objectives, sound data and processes, and governance that makes it safe to run.
That's what separates AI that improves the operation from AI that scales its inconsistencies, and it's where we'll start.
Our approach
Prepare. Select. Design. Implement. Optimise.
- 01
Prepare
Set objectives, assess your data, processes and management information (MI), and build a shared understanding of AI across operations, compliance and leadership. - 02
Select
Choose the use cases worth pursuing, then evaluate partners on capability, governance, operational fit, explainability and regulatory alignment. - 03
Design
Set decision ownership, controls and monitoring, so AI is safe and accountable in live collections. - 04
Implement
Deploy with your teams and partners, so AI works inside the operation rather than beside it. - 05
Optimise
Monitor performance, customer outcomes and model behaviour, and refine so the benefits hold.
What we do
Where we help.
AI readiness assessment
An honest baseline across objectives, data, process, MI and awareness, with a clear next step.
Strategy and roadmap
Where AI will create value in your operation, with use cases, benefits and dependencies in sequence.
Partner selection
An independent, structured selection: use case definition, market review, demonstrations, scoring and commercial support through to contract.
Governance and operating model
Decision ownership, model risk and approval controls, explainability, audit and regulatory reporting.
Implementation support
Hands-on help putting AI into live operations. Typical uses include contact strategy, vulnerability identification, affordability and arrangement setting, quality assurance and agent support.
Explainable analytics
Models from the arum IQ score set, including the Financial Vulnerability Score, Litigation Selection Score and propensity models, with reason codes and auditable decisions.
Proof
47 AI providers, assessed.
We ran a global AI evaluation programme covering 47 providers. Each was assessed on capability, governance, operational fit, explainability and regulatory alignment, with agentic AI capability reviewed as part of the evaluation.
A preferred implementation partner was selected, and the operating model to run it was designed alongside.
We've also defined and prioritised collections AI use cases across contact strategy, vulnerability, affordability, quality assurance and productivity. Each one was sized for value, feasibility and regulatory exposure.
47
AI providers evaluated worldwide
Case study
An AI and automation roadmap for a global debt purchaser.
A global debt purchaser had already invested in conversational AI, voice automation and speech analytics. We assessed the strategy independently and produced a prioritised automation heatmap and delivery roadmap across AI and robotic process automation (RPA).
Global debt purchaserRoadmapA prioritised automation heatmap and delivery roadmap across AI and RPA
Read the case studyWhy arum
Collections experts who have taken AI live.
Independent
We don't resell AI technology, so our recommendations reflect your operation, not a partner arrangement.
We've done it before
The people advising you have run collections operations themselves.
Governance by design
Explainable models, reason codes and auditable decisions are built in from the start.
FAQs
Questions about AI
Is AI ready for collections?
Yes, selectively. The technology is capable, but value depends on preparation. Where objectives are clear, data and processes are sound and governance is in place, AI is already improving outcomes in live collections.
Where do we start?
With a readiness assessment across objectives, data, process consistency, MI and awareness. It shows where AI will deliver, and stops you deploying models on foundations that won't hold them.
How do you keep AI compliant?
Compliance is a design constraint, not a final step. We build in explainable models, reason codes, auditable decisions, clear ownership and ongoing monitoring, in line with regulatory expectations and your model risk policy.
Do you sell AI products or software?
No. We assess readiness, define strategy, select partners, design governance and support implementation. Our recommendations are made in your interest.
Related services
Data and analytics
The scores and models AI decisions rely on, including the Financial Vulnerability Score.
Procure
Independent selection of the collections technology AI has to work with.
Implement
Delivery support to get new capability live.
Advise
Collections strategy and the case for change.
Further reading
- Digital and AI
Let's talk.
Find the AI use cases worth doing. Thirty minutes with one of our senior experts who assesses, selects and deploys AI in live collections operations.
- AtlantaAtlanta Financial Center, 3343 Peachtree Rd NE, Suite 145, Atlanta, GA 30326
- London1 Lyric Square, London W6 0NB
- Sydney97-99 Bathurst St, Suite 1087, Ground Floor, Sydney NSW 2000