AI in Collections, Getting Started: Where It Fits in an Agency's Back Office

If you run a collection agency, you have probably sat through at least one pitch this year promising that AI will transform your operation. Some of those pitches describe real tools. Some describe a demo. Very few start where an agency leader actually has to start: with the question of which work is safe to hand a machine in the first place.

This series is for owners and operations leaders who want a sober on-ramp. Part 1 maps the territory. Where does AI fit inside an agency today, where should it stay out for now, and how do you run a first project your compliance team can review with confidence?

This is general marketing information, not legal or compliance advice. Consult your attorney or compliance officer before acting.

Everyone Is Talking About AI. Fewer Are Using It Well.

Adoption numbers depend a lot on who you ask. McKinsey's State of AI survey, published in November 2025, found that 88 percent of respondents said their organizations regularly use AI in at least one business function, yet only about one third had begun scaling it across the enterprise. The U.S. Census Bureau's Business Trends and Outlook Survey gives a more grounded view of the broader economy: between December 2025 and May 2026, roughly 17 to 20 percent of U.S. businesses reported using AI, while the Finance and Insurance sector sat at 33.9 percent as of May 2026.

The takeaway for agency leaders: financial services firms are ahead of the average business, and most organizations of every size are still experimenting rather than running AI as part of daily operations. If your agency has not deployed anything yet, you are not behind. You are in the majority, and you get to learn from other people's pilots.

Why Collections Has to Start Differently

Most industries can try AI the way they try any new software: pick a use case, buy a seat, see what happens. Collections does not get that luxury, for two reasons.

The first is regulation. Agencies operate under the Fair Debt Collection Practices Act, implemented through the CFPB's Regulation F (in effect since November 30, 2021), along with UDAAP, the TCPA, and state licensing requirements. Regulators have generally not treated a new technology as a reason to relax existing obligations, so the safest planning assumption is that the standards your staff work under apply to your software too. Your compliance team is the right place to confirm exactly how.

The second is trust. The CFPB received approximately 387,400 debt collection complaints in 2025, up from approximately 207,800 in 2024, according to its annual reports. In both years, the most common issue consumers selected was attempts to collect a debt they said they did not owe. Whatever is driving that trend, it reinforces something every agency leader already knows: accuracy matters more than speed. A tool that moves faster but multiplies a data error does not save time. It creates more work and more risk.

So the right first question is not "what can AI do?" It is "where can AI help my team without ever speaking for my agency?"

A Simple Map: Back Office First, Consumer-Facing Last

The easiest way to sort AI opportunities in an agency is by distance from the consumer. The further a task sits from any consumer communication, the lower the risk and the better the starting point.

Zone 1: Internal operations (start here). Work that stays inside your walls and gets reviewed by a person before it matters.

• →Intake and triage. Sorting incoming client paperwork, placement files, and internal email into the right queue, flagging missing documents, and summarizing long attachments so a person can act faster.
• →Internal drafting. First drafts of training outlines, staff-facing policy summaries, meeting notes, and project plans. A person edits and owns the final version.
• →Analytics and reporting. Summarizing operational data you already have, spotting trends in workload or turnaround time, and drafting the narrative around performance reports your team has already verified.

Zone 2: Growing the agency (next). This is marketing and business development, where most agencies feel the pinch of a small team.

• →RFP and proposal support. Organizing past responses into a reusable library and drafting first passes of standard sections for your team to tailor.
• →Content and thought leadership. Outlines for articles, webinar decks, and conference presentations that position your agency's expertise with creditor clients.
• →Client reporting polish. Turning verified numbers into clearer summaries for the clients who place accounts with you.

Zone 3: Anything that communicates with consumers (not a getting-started project). Letters, calls, texts, emails, chat, voice tools, and any decision about when or how to contact someone. This is compliance territory first and technology territory second. Some vendors build products for it, and your counsel and compliance team are the right people to evaluate them. We will not cover it in this series, and we do not recommend it as anyone's first AI project.

The map is deliberately conservative. Zones 1 and 2 hold more than enough time savings to prove value, and none of that work puts your license on the line while your team is still learning how the tools behave.

The Guardrails to Set Before the First Pilot

A getting-started program lives or dies on a few decisions made before anyone opens a chat window.

1. →Decide what data the tool can touch. Consumer account information is sensitive. Many agencies start with work that uses no consumer data at all, which still covers a surprising amount of Zone 1 and Zone 2. Before any consumer data goes near an AI tool, confirm with your compliance, security, and legal teams what is permitted, and check what your client contracts say about vendors and data handling.
2. →Use tools your IT and security team has reviewed. Consumer versions of AI chat tools often carry different data terms than business versions. Read the terms, or have someone who reads them for a living do it.
3. →Keep a human owner on every output. AI drafts. A named person reviews, edits, and approves. Nothing leaves the building on the machine's say-so.
4. →Write it down. A one-page AI use policy listing approved tools, approved uses, and prohibited uses gives your team clarity and gives your compliance team something concrete to review against Regulation F and UDAAP.
5. →Bring compliance in at the start, not the end. The fastest pilots are the ones compliance helped scope. The slowest are the ones compliance discovers after the fact.

Your First 30 Days

If you want a concrete starting point, here is a pilot shape that fits most agencies.

• →Week 1: Pick one workflow. Choose a single Zone 1 task your team finds tedious and that uses no consumer data, such as summarizing internal meeting notes or drafting a first-pass RFP section. Write down how long it takes today.
• →Week 2: Set the rules. Approve one tool, draft the one-page policy, and name the person who reviews every output. Walk your compliance lead through the plan.
• →Week 3: Run it. Have two or three people use the tool on real work for that one task. Keep a simple log of time spent, edits needed, and anything that felt off.
• →Week 4: Decide. Compare the log against your baseline. Keep the workflow, adjust it, or drop it. Then pick the next one.

The goal of the first month is not transformation. It is building a habit of careful experimentation that your team, your compliance function, and your clients can trust.

What's Next in This Series

Part 2 goes deeper on Zone 1: how to set up an internal intake and triage workflow, what to measure, and how to document it so the compliance review conversation is quick. Later parts cover using AI to market an agency, writing an AI use policy people will actually follow, and the questions to ask any AI vendor.

The Bottom Line

AI can give a collection agency real time back, but only if it comes in through the right door. Start in the back office, keep a person accountable for every output, bring compliance in early, and treat anything that speaks to consumers as a separate decision for a later day, made with the right experts. Agencies that build that discipline now will be ready to adopt more as the tools and the guidance mature.

Growing an agency and not sure where AI or marketing fits first? Start a conversation with Yayah Creative Co.

---

This is general marketing information, not legal or compliance advice. Consult your attorney or compliance officer before acting. Statistics are cited from their original publishers (McKinsey, U.S. Census Bureau, Consumer Financial Protection Bureau) as of September 11, 2026.