The right operating model is not automation without employees. It is automation with clearly defined human ownership.
By Vanessa Couch-Laguana, Head of RCM, LunaBill
Automation and artificial intelligence are often discussed as technology initiatives. Organizations evaluate vendors, compare capabilities, and calculate how many tasks a tool might complete.
But after nearly three decades in Revenue Cycle operations, I believe the conversation should begin somewhere else:
What problem are we trying to solve for our patients, our employees, and our organization?
Technology should not become the project itself. It should address a clearly defined challenge—whether that is a growing backlog, missed deadlines, inconsistent follow-up, excessive system navigation, limited staff capacity, or an employee spending most of the day on work that does not require their full expertise.
The most valuable automation opportunities are usually tasks that are rules-based, repetitive, prone to human error, dependent on navigating multiple systems, or difficult to complete consistently because of staffing limitations.
In Patient Access and Revenue Cycle operations, four workflows stand out:
- Eligibility and benefits verification
- Prior-authorization submission and status follow-up
- Notices of admission
- Claim-status follow-up
These workflows are repetitive and time-sensitive, but they also contain exceptions that can affect patient care, reimbursement, compliance, and the patient’s financial experience.
Start with the operational problem
Healthcare organizations should resist the urge to begin with a technology demonstration and then search for places to use it.
Instead, leaders should begin with the work itself.
Where are employees losing time? Where are patients experiencing delays or confusion? Where are deadlines being missed? Where is the organization carrying avoidable financial risk?
Is a staffing shortage creating a backlog? Are employees spending hours navigating payer portals or making routine status calls? Are authorization and notification deadlines being missed? Are eligibility problems being discovered after the patient arrives—or after care has already been provided? Are claims sitting untouched because the team cannot cover the entire inventory?
These are operational problems that automation may help address.
The 2024 CAQH Index identifies eligibility and benefits verification, prior authorization, and claim-status inquiries as significant opportunities for greater electronic automation and administrative efficiency. The potential benefit is not limited to financial savings. Automation can help complete routine work consistently, identify problems earlier, expand account coverage, and allow experienced employees to focus on work that requires judgment.
For leaders, the business case should be stated plainly:
- What work will be completed more consistently?
- What risk will be reduced?
- What capacity will be created?
- What will employees be able to do that they cannot do today?
- How will the patient experience improve?
- How will success be measured?
If those questions cannot be answered, the organization may be evaluating technology before it has defined the need.
Bots, AI, and employees have different roles
Organizations should avoid using “automation,” “bots,” and “AI” as if they all mean the same thing.
In practical Revenue Cycle terms:
- Bots follow predefined rules and complete repetitive steps, such as navigating systems, retrieving information, entering data, and documenting routine results.
- AI tools can help organize information, recognize patterns, summarize results, and prioritize accounts or exceptions for review.
- Employees validate exceptions, apply experience and judgment, resolve discrepancies, communicate when sensitivity or explanation is required, and approve consequential next steps.
The technology should perform work that can be completed safely and consistently within defined rules. Employees should remain responsible for situations that fall outside those rules or could affect care, reimbursement, compliance, or the patient’s financial experience.
This distinction matters because the goal is not to make every interaction automated. The goal is to make sure the right work is handled by the right resource.
A bot may retrieve a benefit response. An employee determines whether that response makes sense for the patient in front of them.
An automated tool may identify that an authorization is pending. An employee determines whether the delay could affect the scheduled procedure.
A system may return a claim status. An experienced Revenue Cycle professional determines what action is needed next.
Four workflows where automation can create immediate value
1. Eligibility and benefits verification
Automated tools can verify coverage, retrieve benefit information, and document routine results before the date of service. This can help teams identify inactive coverage, benefit limitations, plan changes, and missing information earlier.
That early information can make a meaningful difference. It may prevent a patient from arriving for care without active coverage. It may give staff time to correct an insurance record. It may allow the organization to provide a more accurate estimate and a clearer financial conversation.
But an electronic response is not always the end of the process.
Human review should be required when coverage information conflicts, coordination-of-benefits questions arise, the returned plan does not match the patient’s record, or the response is incomplete. An employee must then determine whether the payer or patient needs to be contacted and what information must be corrected.
Automation can gather and organize the information. Employees remain responsible for resolving discrepancies that could affect the patient’s access, estimate, or bill.
2. Prior-authorization submission and status follow-up
Prior authorization often requires employees to move among scheduling systems, electronic health records, payer portals, fax platforms, and workqueues. The work is repetitive, but the consequences of an error can be significant.
Automation can support request submission, routine status checks, deadline tracking, and documentation. It can also help surface requests that need attention before a deadline is missed.
CAQH estimates that adopting the electronic prior-authorization standard could save medical providers and staff approximately 14 minutes per authorization. The same report emphasizes that electronic capabilities should be implemented alongside enhanced workflows.
Human review remains necessary when:
- Clinical documentation is missing
- The payer requests additional information
- A request is partially approved or denied
- The scheduled procedure changes
- The authorized service does not match the scheduled service
- The service performed may not match what will ultimately be billed
- An escalation could affect whether or when care proceeds
These are not simply status updates. They are decisions that may require operational, clinical, or payer-specific judgment.
The human role is especially important when a patient’s care could be delayed or when a provider needs a clear explanation of what is required. Automation can help the team move faster, but it should not make a consequential decision without appropriate review.
3. Notices of admission
Notices of admission are often governed by strict payer timelines. When staffing is limited or responsibility is unclear, a missed notification can create avoidable financial risk.
Automation can transmit notifications, track responses, identify missing acknowledgments, and place results into the appropriate workflow. Employees should review transmission failures, rejected notices, late notifications, and discrepancies involving admission dates, patient status, level of care, or payer requirements.
The objective is not simply to transmit more notifications. It is to ensure that the correct notification is submitted on time, successfully received, accurately documented, and acted upon when an exception occurs.
This is also an area where clear ownership matters. If a notification fails at 4:00 p.m., someone must know who is responsible for reviewing it, how quickly it must be addressed, and what escalation path to follow.
4. Claim-status follow-up
Claim-status work can consume significant staff time. Employees may spend hours logging into portals, waiting on payer phone lines, retrieving routine information, and documenting accounts individually.
That time is not always spent on the claims with the greatest financial or operational impact.
Automation can perform repetitive inquiries and return structured results. Employees can then concentrate on claims that require action, such as denials, unexpected payment outcomes, conflicting payer responses, requests for additional information, or claims that are not progressing within expected timeframes.
CAQH estimates that automating claim-status inquiries could save medical providers and staff as much as 18 minutes per patient visit by reducing manual phone calls and administrative work.
The employee’s role then shifts from repeatedly asking, “What is the status?” to determining, “What needs to happen next?”
That shift is important. It moves employees away from low-value information gathering and toward resolution, prioritization, payer strategy, and communication.
Human review must be a required workflow step
Human oversight cannot be an informal safety net added after implementation. It must be designed into the workflow from the beginning.
When an automated result is incomplete, conflicting, unusual, or potentially consequential, the account should stop and move to human review before any next step occurs.
This requirement should apply to:
- Eligibility conflicts or incomplete benefit information
- Clinical authorization questions
- Changes to scheduled or authorized services
- Notification failures or timing concerns
- Unexpected claim denials or payment outcomes
- Conflicting payer responses
- Requests for additional clinical or financial information
- Situations requiring sensitive patient, provider, or payer communication
Not every communication must be performed manually. Automated tools may complete routine interactions and retrieve information. People should take over when the conversation requires explanation, negotiation, clinical or financial judgment, or sensitivity.
A patient should not have to explain the same insurance problem repeatedly because an automated process cannot resolve it. A provider should not have to search multiple systems to understand why an authorization is stalled. An employee should not be expected to trust an automated result that cannot be explained or verified.
Human review is not a sign that automation has failed. It is a control that makes automation safer and more useful.
This approach is consistent with the broader principles in the National Institute of Standards and Technology’s voluntary AI Risk Management Framework, which encourages organizations to incorporate trustworthiness and risk considerations throughout the design, development, use, and evaluation of AI systems.
Automation should support employees, not devalue their expertise
Employees may understandably worry that automation is being introduced to eliminate positions.
Leaders need to communicate a clearer purpose—and demonstrate it through the design of the work.
Automation should:
- Reduce repetitive work and payer-system navigation
- Increase capacity when staffing is limited
- Improve consistency and accuracy
- Expand the number of accounts the organization can address
- Allow employees to focus on complex accounts
- Create more time for patient communication and support
- Reduce manual touches and avoidable rework
- Give employees better information before they make decisions
- Reduce the frustration associated with repetitive, low-value tasks
The employee’s expertise becomes more—not less—important when routine work is automated. Someone must understand whether the result makes sense, recognize when the situation falls outside the standard rules, and determine what should happen next.
Technology can retrieve a claim status. It takes Revenue Cycle knowledge to understand why the claim is not progressing and what intervention is required.
Technology can identify a coverage discrepancy. It takes a trained employee to determine whether the patient needs to be contacted, whether the payer record must be corrected, and how to explain the issue respectfully.
That is the human value leaders should protect and expand.
Redesign roles and handoffs before implementation
Two of the greatest mistakes organizations make are implementing technology without redesigning roles and handoffs, and failing to define when human review is required.
If an automated process identifies an exception but no one owns it, the work has not been improved. It has simply been moved.
Before implementation, leaders should define:
- Who owns each automated workflow
- Who reviews each category of exception
- How quickly the review must occur
- Where results must be documented
- How accounts move between automated and human work
- Which conditions require escalation
- Who receives each escalation
- What authority the reviewer has to act
- How unresolved issues move to the next level
- What measures will determine whether the workflow is successful
- What happens when the technology is unavailable
- How employees can report inaccurate or unsafe results
Ownership, escalation rules, timing, expectations, and performance measures must be established before launch.
Leaders should also ask whether the new process creates work in another department. A faster eligibility response is not an improvement if it creates a larger unresolved exception queue. Automated claim-status results are not valuable if no team has the capacity or authority to act on them.
The 2025 CAQH report on FHIR, interoperability, and electronic prior authorization identifies staffing and internal resources, workflow integration, vendor readiness, and payer implementation as factors affecting provider adoption. This reinforces an important lesson: technical capability alone does not create an effective operational process.
Frontline employees should help build and test the process
Employees who perform the work should be included in discussions about the current workflow and its pain points. They often know where duplicate steps occur, where information is difficult to locate, and where the documented process differs from what happens in practice.
They also know which exceptions are truly urgent and which ones can wait. That practical knowledge is essential.
Frontline employees—or designated representatives from each affected team—should participate in testing. They can help determine whether:
- The technology follows the actual workflow
- The information returned is accurate and usable
- Results appear in the correct system and field
- Exceptions are routed to the appropriate person
- Employees can understand why an account was escalated
- The new process creates unintended work elsewhere
- Documentation supports downstream teams
- Patients and providers receive clearer, more timely information
- The process works under realistic conditions, including incomplete data and system downtime
Testing should include ordinary cases, difficult cases, and failure cases. A workflow that works only when every data field is complete is not ready for production.
However, frontline employees should not be solely responsible for defining every exception and control. Human-review requirements should be established collaboratively by Patient Access and Revenue Cycle leaders, frontline representatives, clinical leaders when appropriate, compliance and privacy teams, IT and system experts, and the technology vendor.
Final approval should come from a vice president or higher-level senior leader. Executive sign-off should be more than a formality. Senior leaders should understand what was automated, how roles and handoffs changed, which risks were identified, when human review is required, who owns escalations, and how success will be measured.
Governance cannot end at go-live
Automation cannot become a “set it and forget it” process. Payer rules, portal designs, authorization requirements, workflows, and system configurations can change.
A process that was accurate in January may not be accurate in June if a payer changes its portal, documentation requirements, response language, or submission rules.
Ongoing governance should include:
- Appropriate access to patient and payer information
- Privacy and security requirements
- Audit trails showing what the technology completed
- Documentation of automated and human actions
- Routine accuracy testing
- Change controls for payer, portal, and workflow updates
- Procedures for failed transactions and system downtime
- Defined owners for monitoring and correcting performance issues
- Scheduled post-implementation reviews
- A clear process for employees to report concerns and recommend improvements
The organization should also be able to trace how a result was obtained, why an account was routed for review, who made the next decision, and what action was taken.
Governance should include a regular review of exceptions, not just overall volume. Exceptions often reveal where the rules are incomplete, where the data is unreliable, or where the workflow needs to be redesigned.
Measure outcomes, not just automated volume
The number of transactions completed by a bot does not demonstrate whether the workflow is successful.
Leaders want to know whether the investment improved performance, reduced risk, supported employees, and made the patient experience better.
Leaders should evaluate measures that show operational, financial, workforce, and patient impact, including:
- Percentage of accounts requiring human review
- Eligibility-related denials
- Authorization-related denials
- Missed authorization or notification deadlines
- Claim-status productivity
- Percentage of the account inventory receiving timely follow-up
- Reduction in manual touches
- Reduction in rework
- Employee capacity
- Employee turnover or engagement, when appropriate
- Patient wait times
- Exception-resolution time
- Validated accuracy rate
- Net financial impact
- Return on investment
- Patient and provider complaints related to the workflow
These measures should be reviewed together.
A declining human-review rate may appear positive, but not if exceptions are being overlooked. Higher claim-status volume has limited value if employees cannot act on the returned information. Faster eligibility verification does not help if discrepancies remain unresolved before the date of service.
Organizations should compare results with a documented preimplementation baseline and conduct scheduled reviews after launch. The purpose is to determine whether the technology is solving the original problem—not simply whether it is operating.
A useful executive dashboard should answer five questions:
- Is the process faster?
- Is it more accurate?
- Is it reducing financial or compliance risk?
- Are employees spending more time on higher-value work?
- Are patients and providers experiencing fewer delays and clearer communication?
An automation-readiness test for leaders
Before automating a workflow, leaders should be able to answer:
- What operational problem are we solving?
- Which parts of the workflow are rules-based and appropriate for automation?
- Who owns the automated results?
- What conditions stop the workflow for human review?
- How quickly must exceptions be resolved?
- Where will automated results and human decisions be documented?
- How will accuracy, timeliness, capacity, rework, and downstream outcomes be measured?
- Have frontline representatives tested the workflow using realistic scenarios?
- Are privacy, security, audit, downtime, and change-management controls defined?
- What happens when the technology produces an incomplete, inaccurate, or unexpected result?
- How will employees and patients receive help when the automated process cannot resolve an issue?
- Has a vice president or higher-level senior leader reviewed and approved the final workflow?
If these questions cannot be answered, the organization may not be ready to automate that process.
Technology must remain connected to the work
The strongest automation strategy is not the one with the most advanced terminology or the greatest number of automated tasks. It is the one that solves a meaningful operational problem and improves the work surrounding it.
Technology succeeds only when roles, handoffs, timing, and accountability are clear.
It also succeeds when employees trust the process, understand their role, and believe the technology is helping them serve patients and providers more effectively.
Automation can help Revenue Cycle teams manage staffing limitations, reduce repetitive work, meet critical deadlines, cover more accounts, and identify problems earlier. But it cannot replace operational ownership or the experience employees bring to complicated decisions.
The goal is not to remove people from the Revenue Cycle. It is to use technology to direct their time and expertise toward the work where people create the greatest value.
For patients, that may mean fewer surprises, fewer delays, and clearer answers.
For employees, it may mean less time spent searching, clicking, waiting, and reworking the same account.
For leaders, it should mean measurable improvement in capacity, performance, risk, and experience.
Technology should solve the operational problem. It should never become the project itself.
