Four things we are hearing and what they signal for the work ahead.
The first half of the year put us in a lot of rooms. Four conferences, a stack of panels and webinars, and a steady run of client conversations with creditors, agencies, and the operators doing the work.
Different audiences and agendas, but the same problems surfaced everywhere. The most useful exchanges happened off the main stage, in the roundtables, the hallways, and the calls that came after, where people say what’s actually working and what isn’t.
Nobody was short on ideas. Everyone was short on bandwidth. Across all of it, one thing was clear. For most teams, what stands between them and working AI isn’t the technology. The technology is ready and the ROI is increasingly clear. What keeps the work from shipping is everything around it: readiness, alignment, attention, and capacity.
Getting from a decision to live production runs through four stages. The work can stall at any of them.
Here’s what we heard.
The “ready-to-go” product still took 18 months
“It took 18 months to implement an out-of-the-box solution.”
We heard a version of this more than once. Creditors and agencies are buying AI products marketed as ready to deploy, then discovering production is still a long way off. Integrations break. Data needs cleanup before the model is usable. Legal and compliance reviews stack up fast.
The pattern underneath it: data readiness isn’t the same as AI readiness. The bottleneck is the path from decision to production, and collections operations don’t have an 18-month runway. The portfolios don’t wait.
The teams making progress started with one well-chosen problem, proved value fast, and built the plan forward from evidence instead of assumptions.
Read more: What we learned the hard way about putting AI to work
Internal teams aren’t speaking the same language on AI
“Our teams aren’t aligned on AI language or prioritization.”
Operations wants throughput. Compliance wants guardrails. IT is already at capacity. Leadership wants the strategy before the spend. Four groups, four definitions of what a good AI decision looks like.
When those groups share a common language, decisions accelerate and the right work gets prioritized. When they don’t, every use case gets re-litigated from scratch.
Compliance belongs in this conversation early. The regulatory environment in collections isn’t getting simpler. Organizations treating compliance as a cost center are falling behind, and the ones building it as an operational capability are pulling ahead. Done right, AI makes compliance easier to document and defend in an audit, and that documented compliance is increasingly what wins business and keeps it.
What helps is a shared prioritization framework that scores use cases on business impact, feasibility, compliance sensitivity, and time to value. Build that first. Then build the solution.
Email is still the workflow, and it needs AI
“Email is still how our business runs. We need automation there.”
Dispute intake, complaint handling, consumer correspondence, client comms. It all flows through email, and it’s still largely manual and hard to scale. While AI investment chases newer channels, email remains the backbone of daily work for most creditors and agencies.
The technology to fix this has existed for years. What’s been missing is the decision to make email a priority. The volume is high, the patterns repeat, and the compliance stakes are real, which means the ROI is real too.
Working AI on email in collections looks like three things: every inbound email read, categorized, and prioritized; the right work routed to the right person with context attached; and a drafted response based on policy and history that a person reviews before it sends.
Email automation is a known pattern with a known payoff, and it’s still sitting on most roadmaps as “someday.” The teams that move it to “now” recover capacity they didn’t know they had.
The backlog is bigger than the team
“We’re underwater. BAU owns the calendar. AI keeps getting pushed.”
Collections teams aren’t short on ideas. Better dispute routing, automated complaint coding, smarter omni-channel orchestration, digital self-service. The list is long. The problem is that today’s urgent work continually crowds out the work that would reduce tomorrow’s backlog.
Add budget pressure to that. Economic uncertainty is real, leadership isn’t approving open-ended investment, and revenue pressure makes it worse. The instinct is to wait. But waiting has a cost. Every quarter without better automation is a quarter of manual effort, compliance exposure, and missed recovery that doesn’t come back.
This is where prioritization stops being a planning exercise and becomes the actual work. When capacity is tight and budget is tighter, what you choose not to build matters as much as what you do. Not everything on the backlog is equal. Use cases at the intersection of high volume, regulatory sensitivity, and repeatable work almost always deliver the best ROI. Identify the two or three with the highest impact, lowest complexity, and fastest path to measurable return. Start there. Prove value. Use the result to fund the next one.
Dispute-first requirements have recently taken effect, and the regulatory load isn’t easing. At the scale this industry runs on, with billions in annual placements, even a small lift in liquidation rates is material. That is why the teams that move now pull ahead of the ones that wait.
What it looks like when teams pick one problem
Four examples, each one a collections team that picked a single problem, proved value fast, and built from there.
So what is the problem?
If it was never the technology, it’s the path from decision to production. Bandwidth, alignment, and the discipline to pick one thing and finish it before starting the next.
If there’s one lesson from these conversations, it’s that progress rarely comes from the biggest initiative. It comes from the right first one. The teams making progress aren’t the ones with the largest AI budgets. They’re the ones disciplined enough to solve one meaningful problem, prove value, and build momentum from there.
We’ve spent 28 years in financial services, embedded across the lifecycle from originations to collections, working with banks, credit unions, agencies, debt buyers, and fintechs. We understand the regulations, and we’ve seen what it looks like when a backlog grows so large that credit bureau reporting gets suppressed just to manage compliance risk. We’ve built the fixes that turned that around.
We don’t sell transformation. We help you figure out what to build first, build it fast, and make the value visible before you commit to the next step. If that’s the conversation you’re having internally, it’s the one we’re built for.
Ready to move forward with intention? Let’s talk.
Related reading: The real AI risk in financial services isn’t the technology
SingleStone · Engineering for the human experience.®