According to CareMetx's 2026 Patient Services Trends Report, 94% of patient services leaders are either using or actively planning to adopt AI point solutions. AI and predictive analytics ranked as the #1 technology expected to drive transformation over the next 12 months, outpacing every other capability on the list.
That level of consensus is significant. It reflects genuine investment, real momentum, and a shared conviction that AI belongs at the center of how patient access operates going forward.
And yet, across the same survey, the picture that emerges is one of persistent friction: fragmented data, limited interoperability, and manual handoffs that constrain what automation can actually deliver. Programs that have adopted AI at the step level are still struggling to produce consistent, reliable outcomes across the full patient journey.
Understanding why that gap exists matters both for how the industry deploys AI today and for what it should be building toward.
The Point-Solution Problem
Across the patient access journey — intake, benefit verification, prior authorization, affordability support, dispense and ongoing care — there is now an AI or point solution available for nearly every discrete activity. Adoption follows a recognizable pattern: start with high-burden administrative functions, then extend into access-critical areas as confidence grows.
That approach is rational; automating benefit verification, accelerating prior authorization, and streamlining intake are all meaningful improvements that reduce friction and create real value. But that value is often confined to the step where it’s created, and doesn’t consistently carry forward across the broader process.
When a benefit verification is completed accurately and quickly, that information still has to move to the provider, to the PA team, to the pharmacy. And in most programs today, those handoffs are where things begin to break down. Data arrives in non-standard formats, updates don’t flow to the right systems, and actions that should be triggered automatically end up requiring manual follow-up instead.
The report puts it plainly: programs that layer AI onto fragmented or manual workflows are unlikely to see sustained returns, because structural constraints continue to limit impact regardless of what any individual tool can do.
Download the 2026 Patient Services Trends Report →
The Interoperability Gap Is Real, Even When Readiness Feels High
One of the more telling findings in this year's report involves the gap between how prepared organizations believe they are, and where they realistically stand in terms of interoperability.
More than half of respondents describe themselves as "very prepared" for interoperability and data sharing requirements. But when asked to characterize the state of their programs, the majority report being in planning or early implementation, with relatively few nearing completion.
That gap has direct implications for AI specifically. The performance of any AI-driven capability is not just a function of the model itself. It depends on data arriving in the right format, at the right moment, in a context where it can actually trigger action. When those conditions aren't met, AI doesn't fail visibly, but it operates in a narrower lane than it was designed for, doing less than it could without anyone necessarily knowing what's being left on the table.
More than 40% of survey respondents also cited issues with data format and actionability — information arriving too late, structured in ways that cannot feed downstream workflows, or surfaced in a system where it has no practical use at that point in the journey. These cases represent a class of infrastructure problem that sits beneath the surface of most AI deployments, and quietly limits their reach.
What a Stronger Approach Looks Like
In a more connected model, signals generated at each step — coverage determinations, payer requests, authorization outcomes, dispensing activity — don't stay contained within the workflow where they originated. They carry forward, informing what happens next. Each stakeholder acts with the context of what has already occurred, rather than starting from a partial picture and reconstructing what they need.
Over time, that connected flow of information does more than improve individual handoffs. Patterns emerge. The system begins to anticipate where delays are likely, how specific payers behave, which interventions are most effective for particular patient populations. The result is not just faster execution at discrete points — it is more coordinated, more predictable access across the full journey.
This is what Collective Intelligence℠ is designed to support. Not AI applied to isolated steps, but AI used to connect them — so that information moves with the patient, decisions are made with full context, and the access journey holds together from enrollment through adherence.
The Question Worth Asking
If nearly every program in the industry is adopting AI, and most are still wrestling with fragmentation, the strategic question is no longer whether to invest in AI. It is whether the infrastructure exists to let AI work across the journey, not just within it.
The 2026 Patient Services Trends Report offers a detailed look at where the industry stands on that question, including how programs are structured, where technology is being deployed, and what constraints are limiting the return on those investments.