SENA Insights

Can Your Front End Actually Make Something Happen?

Posted on November 5, 2026

← Back to Blog

Can Your Front End Actually Make Something Happen?

SENA Health November 5, 2026

By Anthony J. Wehbe, DO, MBA, FACOI | Founder & CEO, SENA Health | Practicing Internist, Senacare | Former Health-System Physician Executive

Why the next frontier in access is moving from understanding intent to resolving the need

Healthcare is getting much better at understanding what patients are asking for.

AI can answer a call, interpret a portal message, identify intent, gather information, summarize a problem, and increasingly complete parts of a transaction. That is real progress, and it will matter a great deal.

But from where I sit — still seeing patients most weeks, still running a technology company built to fix this exact problem — understanding the request is only part of the job.

Not long ago I wrote a referral by hand from a patient’s kitchen table, for a home-bound man in his nineties with new chest discomfort. I didn’t trust our own queue that day, so I faxed it myself and called the cardiology office forty-eight hours later to confirm it had arrived. It hadn’t. Nothing was wrong with anyone’s intentions. Everyone involved — me, the office staff, the answering service — did their piece of the job correctly, and the referral still sat in nobody’s queue for two days. That is the problem I want to talk about.

The harder question is whether the system can actually do something useful with what it now understands.

If a patient says, “I need to see someone about this chest discomfort,” recognizing the request is not enough. The system has to know whether it can proceed through an established workflow, whether a clinician needs to review the issue, whether the patient should be routed somewhere else, and whether that next step actually happens.

That is where front-end performance moves from communication to execution. And that is where many current workflows still break down.

A scheduling team may complete its task, but the patient still does not have the right appointment. A referral may be entered, but no one knows whether it was received or acted on. A medication request may be routed correctly, but it sits until someone else picks it up.

In practice, I see this constantly. The problem is rarely that nobody touched the request. The problem is that too many people touched it, and nobody truly owned the path to resolution.

Everyone can do their part, and the patient can still be stuck.

One Accountable Owner at Every Step

I learned this the hard way at Kennedy Health System, when our sepsis mortality rate ran close to the New Jersey average of roughly 50 percent. The instinctive fix was more nursing coverage — more frequent vital-sign checks, in other words, more staff watching more closely. That helped at the margins, but what actually moved mortality below 10 percent, and made us the only hospital in the country recognized that year with the National Sepsis Hero Award, was building one workflow that owned escalation from the first abnormal vital sign through antibiotic administration, with a named, accountable owner at every step. We didn’t solve that by adding people to watch harder. We solved it by making sure the hand-off couldn’t get dropped without someone noticing immediately.

A high-performing front end cannot control every clinical outcome, and it cannot manufacture capacity that does not exist. But it can do a much better job of executing against what is available.

That means seeing the appointment that is actually open, applying the right rules, moving the patient to the right next step, recognizing when the normal workflow does not fit, and getting the issue to the right person when clinical or operational judgment is required.

Most importantly, it means knowing whether what was supposed to happen actually happened.

That is the execution challenge.

Design Around the Work

At SENA Health, that is why we start with workflow rather than technology. We look at what the patient is trying to accomplish, what information is needed, where the request should go, where it usually stalls, and what can be handled automatically versus what requires trained staff or clinical judgment. Then we design around the work.

Sometimes AI should handle most of the task. Sometimes a trained access coordinator needs to step in. Sometimes the system needs to stop and escalate because the issue has crossed into clinical decision-making.

The goal is not to automate as much as possible. The goal is to make the workflow perform from end to end.

As AI gets better at understanding the request, the chart, and the likely next step, the limiting factor increasingly becomes whether the organization can act on that information.

That is why I’d push back on the way a lot of boards are currently evaluating access technology. An AI that understands intent brilliantly and then hands off into a workflow with no accountable owner has not solved your access problem — it has made your organization faster at failing to resolve it. I would rather run a slower workflow that reliably closes the loop than a fast one that reliably drops it. Automating an unowned hand-off doesn’t create accountability. It just moves the drop earlier in the process, to a point where it’s harder to see and further from where anyone will catch it.

Measure Whether the Patient Moved Forward

It also changes how leaders should think about measuring front-end performance. Calls answered, abandonment rates, response times, tasks completed, and messages closed still matter. But they tell us how well the front end performed a task. They do not tell us whether the patient actually moved forward.

Did the patient get the appointment they needed? Did a referral loop actually close? Was a medication issue resolved, and were concerns routed to a clinician when clinical judgment was required? Those are harder measures, and they are much closer to what patients and physicians actually experience.

We are trying to make the front end of the enterprise more responsive. Responsiveness without follow-through can simply create faster hand-offs into the same fragmented system.

I keep coming back to the same five levers, because they are the only ones that tell me whether an access investment is actually working: capacity utilization, leakage, clinician burden, patient retention, and margin. Execution failures hide inside every one of them, and they are the hardest failures to see, because most dashboards only measure activity — calls answered, referrals entered, messages closed — not resolution.

That’s part of why, when we went live with our $120 million Epic implementation across Jefferson Health New Jersey, I cared less about go-live speed than about whether the system could actually tell us, downstream, if a referral closed or an order was acted on; we ended up ranked in the top quartile nationally on Epic’s own scorecard, and that visibility is what let us catch the hand-offs that were failing instead of assuming they were fine because a task had been marked complete.

That is the number I want every physician enterprise asking an AI vendor to explain: not how many interactions did your tool handle, but how much leakage, burden, and cost did it actually remove — and how do you know?

I believe the next opportunity is to make the front end more accountable for moving the patient toward the right next step, with fewer dead ends, fewer callbacks, and fewer unresolved tasks. That is not simply an AI problem, and it is not simply a staffing problem. It is an operating problem — which is exactly why SENA Health starts with workflow rather than technology or staffing.

Executive Checklist

Leadership should be able to answer four questions:

  1. Once we know what the patient needs, can we actually do something about it?
  2. Where does the patient still get stuck, even though everyone did their part?
  3. What can technology handle, and when does a person or clinician need to step in?
  4. Are we measuring the work we completed, or whether the patient actually got where they needed to go — and what that gap is costing us?

The front end cannot control every outcome, but it can take much greater responsibility for getting the patient to the next right place. That is where understanding becomes useful, where navigation becomes execution, and where the front end starts to perform like an operating capability rather than a collection of transactions.


Want to talk through your organization’s front end? Schedule a conversation with Dr. Wehbe.