WizKids Dental, the first Aria deployment.

WizKids Dental is the first live deployment of the proprietary in-house engineered Velzyx Aria voice runtime. The practice is in Southern California. The system has been running long enough to be worth writing about. This is the account of what the deployment delivers and what has been learned operating it.

A note on what this page is and is not. The study is written with the practice's permission and reviewed by the practice owner before publishing. It contains no patient data, no identifying details about anyone who has called the practice, and no information that crosses the privacy line we drew with the practice during the build. It also contains no fabricated metrics. We are not going to tell you the system saved a percentage of staff time we did not measure, or recovered an amount of revenue we cannot prove. We are going to tell you what the system does, how we built it, and what is different inside the practice with the system in production.

The practice

WizKids Dental is an appointment-driven Southern California practice with a multi-provider mix and a substantial new-patient inflow from local search. The practice owner and the front-desk team are good at what they do. The constraint on the operation, when we first spoke, was not the quality of the team but the math of the team's day. Every front desk on the planet, in every appointment-driven business we have studied, eventually hits a ceiling where the work coming in over the phone exceeds the throughput of the humans available to take it. WizKids was running into that ceiling regularly.

The owner had tried generic answering services. The owner had tried a horizontal voice assistant pitched as a one-size-fits-all dental tool. Neither stuck. The answering service handled call volume but had no context for the practice's actual workflow. The horizontal assistant had context for "dental" in the abstract but no context for WizKids in particular. By the time we met, the owner had landed on a clear ask: a system that handled the practice's calls the way the practice handled its calls, and got out of the way fast for anything outside that.

This is the brief Velzyx exists to take. WizKids became the first live deployment of the proprietary Aria voice runtime under that brief, and the operator agreed to be the first case study in this collection because the deployment surfaced patterns worth being public about.

The WizKids operation

What the front desk actually deals with is not in any document. Calls coming in while a check-in is in progress. Insurance questions that require the front desk to pull a specific carrier's coverage rules from memory. A parent calling about a child whose coverage is on the other parent's plan. A long-time patient calling to move an appointment that was scheduled six months ago, and the front-desk team holding the context for that patient in their head while answering. A new patient calling at 11:40, audibly shopping around, and asking three or four questions in sequence that the front desk has to answer cleanly to win the call.

The deployment is configured for that reality. The providers, the operatory layout, the new-patient flow the owner wanted preserved, the insurance plans the practice is in-network with, the language the front desk uses when discussing cost, and the narrow set of cases the owner explicitly wanted routed to a human — clinical conversations, financial disputes, anything emotional. Every other inbound call is scoped for Aria. Aria is the front desk on the phone.

What Aria owns at WizKids

The Velzyx-engineered system at WizKids is configured as the front desk on the phone — first point of contact for every inbound call — and escalates the narrow set of clinical, financial, and emotional cases to a human path the practice already runs.

Aria owns the new-patient call, end to end, when the new-patient call falls into the lanes the owner agreed to. It greets the caller in the practice's voice. It asks the intake questions the front desk used to ask, in the order the front desk asked them. It captures insurance information against the carriers the practice is in-network with, and it captures the subscriber-and-dependent structure correctly when a child is calling on a parent's plan. It books the new patient into the operatory and the provider that fits the visit type the practice has defined for that intake. The booking writes back into the practice's scheduling system so the front desk sees the appointment the way they would see any other.

Aria also owns hygiene recall outreach. The recall work was the cleanest slice for the system to take, because it is structured, it is repetitive in shape, and it was the slice the human team had been losing the most ground on. Aria reaches out to patients whose recall window has come around, offers appointment options that fit the practice's hygiene block structure, and books the visit when the patient is ready. The owner reviewed the recall script. The owner reviews the live recall traces.

Aria does not own clinical conversations, financial disputes, anything emotional, or any case where a patient is asking a question that the practice owner wants the human team to handle directly. The escalation path is fast and clean. The system identifies the signal, hands the call to a designated human channel, and writes the trace into the practice's record so the human picking up has context.

What the deployment delivers

The Velzyx-engineered system at WizKids is the conversational layer the practice's callers actually talk to, the deterministic logic that handles scheduling and writeback, the integration surface that lands appointments cleanly on the front-desk screen, and the observability surface that makes everything legible to the practice owner. Each layer is the WizKids-specific configuration of the proprietary Velzyx primitives.

The conversational layer behaves like the WizKids workflow document says the front desk behaves — the cases the practice handles get handled the way the practice handles them. The integration writeback delivers appointments with the visit type the practice expects, in the operatory the practice expects, in a form the front-desk team recognizes the moment they look at the screen.

The observability layer is part of the deployment, not an afterthought. Every call Aria takes produces a trace the front-desk team can pull. Every booking the system writes is reviewable. Every escalation is logged. The owner can sit down with the deployment and see what the system did, why it did it, and what the team did with the handoff. Legibility is part of what the operator receives.

Go-live: the first live call

The first call Aria took at WizKids was a new-patient inquiry on a Wednesday afternoon. The call landed inside the lanes the design had drawn, the system asked the intake questions in the order the workflow document specified, the booking wrote cleanly into the scheduling system, and the front-desk team saw the appointment on screen before the patient hung up.

The next call was a current-patient question that the design had decided the human team should handle. The escalation pattern fired the way the design said it would. The human picked up with a trace already in front of them.

The first day was, by design, boring. A production deployment is supposed to be boring on the first day. The work happens before go-live; what shows up afterward is whether the work was done.

What is different inside the practice

The plain-language description of what changed, in the owner's words and the team's words, sits in a small handful of statements.

Every inbound call is answered. Aria is the first voice every caller hears on the practice's existing number — new patients, current patients, recall callers, emergency triage — around the clock. The team knows it. The patients calling at 7pm on a Tuesday know it.

The schedule fills itself in. Hygiene recall outreach, which used to slip when the team had no time for it, runs as a steady background. The patients the practice already worked hard to acquire are not quietly aging out of the schedule.

The practice owner stopped checking voicemail at night. That sentence is the one the owner uses when asked what changed. The shift is not a number. It is a habit the owner used to have, born from anxiety about missed calls, that the owner has put down. That is a real change inside the practice, and it is the kind of change a small operator pays attention to even when it does not show up on a dashboard.

We are deliberately not putting percentages or dollar figures on this page. We could collect numbers and put them up. The numbers would be ours, not the practice's. The qualitative statements above are the practice's own. They are the honest read of what is different. We trust an operator reading this study to weigh those statements as they deserve to be weighed, which is more than a fabricated metric deserves.

What we learned

The deployment surfaced a small number of lessons the studio is carrying forward into every future Aria deployment.

The first is that the human team's trust in the system is built by observability, not by performance claims. The WizKids team trusted Aria after they could pull a trace and see what the system had done. They did not trust it earlier, and they should not have. We engineer observability accordingly across every deployment now.

The second is that the recall slice is one of the highest-value places for the system to take work that the human team has structurally been losing. Recall is structured, repetitive, and the kind of work that a human team will deprioritize when the day gets busy. A system that handles it without deprioritizing is a system that compounds for the practice over time. Every Aria build now treats the recall lane as a first-class scope item.

The third is that the escalation pattern, more than any single piece of the design, determines whether the operator trusts the system in the long run. A system that escalates cleanly earns the territory it is given. A system that hesitates loses it. We tune escalation hard, with the operator, before go-live.

What is next at WizKids

The deployment is live. Velzyx operates the system alongside the practice. Live operation is ongoing, with regular evaluation against the production version, drift monitoring, and engineering work that follows the practice's workflow as it shifts. The owner has the engineer's direct line. There is no account manager in between.

The next layer of work at WizKids is a deeper integration with the practice's existing systems and an expansion of the workflows Aria runs alongside the inbound front-desk role — broader recall outreach, payment link follow-up, and structured patient communications. The owner and the studio agreed at scope time that expansion would be the operator's choice, paced by what the operator was willing to hand over as trust grew. The conversation about the next layer is happening at the time of writing.

If you are weighing a similar deployment

The shape of a Velzyx engagement, including what an operator sees at each milestone, is described in detail on the engagement page. If your operation looks similar to a multi-provider dental practice handling new patients across providers, the industries page for dental is the right next read. The system itself is described on the Aria page. The engineering posture behind every Velzyx-engineered system is on the methodology page.

If you would rather skip the documentation and ask the questions directly, the studio's contact page is at /contact.html. We read every message and respond.