Avoca Systems

Radiology booking is a phone problem before it is a software problem. A patient calls after their GP sends a referral; the line rings, the receptionist is juggling three other calls, and the caller gives up and tries the next clinic on the list. That lost call is not just a missed appointment—it is revenue walking out the door. Avoca Systems, a pre-seed startup from Y Combinator's latest batch, has built an AI layer that sits on top of this reality, not around it. Its pitch is not "automate your scheduling" but "capture every booking across all your channels"—and the distinction matters.

Avoca's platform handles the messy, human end of radiology: answering calls, following up on referrals, sending reminders, and even recalling patients for follow-up scans. It does this by learning how a specific clinic operates, not by forcing a generic workflow. The company calls this "operational intelligence," and it is the core of the product. This article looks at how Avoca is trying to solve a problem that most AI scheduling tools ignore: the complexity of a real radiology network.

Why radiology bookings still fail on the phone

Radiology is a high-volume, high-stakes scheduling environment. A single clinic might handle hundreds of referrals a day, each requiring a slot, a prep instruction, and a patient who actually shows up. The phone is still the primary channel for many patients, especially older ones or those with complex conditions. But phone booking is inefficient: it ties up staff, creates long wait times, and drops calls when demand spikes.

Avoca's website is blunt about the cost: "Every unanswered call and stalled referral is demand that never converts." That is the problem statement. The company's answer is a voice AI agent that can handle complex radiology bookings—not just "press 1 for appointments" but the kind of back-and-forth that involves asking about prep, checking insurance, and finding a time that works. The site claims it handles "the complex radiology bookings other tools pass to a human." That is a strong claim, and it is the crux of Avoca's value proposition.

The phone is not the only channel. Avoca also covers online self-service booking and eReferrals direct from GPs. But the phone is the hardest problem, and it is where Avoca seems to be focusing its AI. The company's customer story features a quote from Dylan Campher, CEO of Partnered Health, who says the AI converted 30 bookings in half a day, roughly the output of two full-time booking agents. That is a concrete, if anecdotal, measure of the problem Avoca is solving.

The 86% call resolution number that matters

Avoca's website prominently displays an 86% call resolution rate. This is a specific, measurable claim, and it is the kind of number that should make a radiology network operator sit up. Call resolution rate—the percentage of calls that end with a completed booking or a clear next step—is a core operational metric for any call center. For radiology, it is even more critical because every unresolved call is a potential lost patient.

The number appears twice on the site, once in a customer story and once in a testimonial. It suggests that Avoca's AI is not just answering calls but actually resolving them at a rate that approaches human performance. The site also shows a 4.6/5 customer satisfaction score, which is high for any automated system. These numbers are early and come from a single customer, but they are the kind of evidence that gives Avoca credibility in a market full of vague AI promises.

It is worth noting that the site does not disclose how many calls the 86% figure is based on, nor how it is measured. But the fact that Avoca is willing to publish a specific metric, rather than a fluffy "improves efficiency," is a signal of confidence. For a pre-seed company, that is a bold move.

Encoding clinic knowledge, not just automating calls

The most interesting part of Avoca's product is not the AI itself but the system that lets clinics teach it their rules. The website shows a screenshot of an "operational intelligence" interface where a clinic can define entries like "Prep required for abdominal ultrasounds" and "Locations & access." These are not generic scheduling parameters; they are the idiosyncratic details that make each radiology clinic different.

Avoca's system learns from recent calls and suggests entries. The screenshot shows a suggestion: "Prep required for abdominal ultrasounds — Based on four calls." This is a form of continuous learning, where the AI picks up patterns from real conversations and turns them into structured rules. The clinic can then approve or edit these rules, building a knowledge base that is specific to its operations.

This is a clever approach because it solves the cold-start problem. A generic AI booking system does not know that a particular clinic requires fasting for a certain scan or that a specific location has limited parking. Avoca's system learns these details from the calls it handles, and the clinic can refine them. This means the AI gets smarter over time, and the clinic retains control.

The site also shows that clinics can define rules for "Locations & access," "Services offered," and "Fees & billing." These are the operational details that trip up generic schedulers. By encoding them, Avoca ensures that the AI does not just book a slot but books the right slot, with the right prep instructions, at the right location, and with the correct billing information.

The RIS integration that makes it invisible

Avoca does not want to replace a radiology network's existing systems; it wants to sit inside them. The website lists integrations with RIS (Radiology Information System) vendors like Pro Medicus, Comrad, and Kestral. This is a critical move. Radiology networks have invested heavily in their RIS, and they will not rip it out for a startup. Avoca's integration-first approach means it can be deployed without disrupting existing workflows.

The site says Avoca "connects directly to your RIS and works within your existing radiology setup." It can run "silently alongside your current workflows, or as your primary interface for patient access across the network." This flexibility is important. Some clinics may want Avoca to handle only the front-end patient communication, while others may want it to be the main booking interface.

By integrating with the RIS, Avoca ensures that bookings are actually secured in the system of record. This is not a standalone scheduling tool that creates a parallel database; it is a layer that writes into the systems the clinic already trusts. This reduces friction and makes it easier for staff to adopt.

Scaling networks without scaling headcount

Avoca's core pitch is that it lets radiology networks grow without adding staff in proportion. The website says: "Growing your network shouldn't mean hiring in proportion." This is the economic argument. A network that adds a new imaging site typically needs to hire receptionists, schedulers, and coordinators. Avoca claims to absorb that workload, allowing the network to scale its footprint without scaling its back office.

The customer story from Jobfit (a Partnered Health brand) illustrates this. The quote from Dylan Campher suggests that the AI's output is equivalent to two full-time employees. For a network with many sites, that multiplier could be significant. If Avoca can maintain an 86% call resolution rate across a network, the savings in headcount could be substantial.

But there is a risk: the AI may not handle every edge case, and a human may still be needed for complex calls. Avoca's system is not designed to replace all staff, but to handle the bulk of routine bookings, freeing humans to focus on "the things that really matter," as Campher puts it. This is a more realistic and honest positioning than "AI replaces your entire front desk."

What Avoca's early customer evidence shows

The website lists several healthcare networks as customers or partners: Lumus Imaging, Partnered Health, Western Radiology, ForHealth, Precision Imaging Partners, and Ochre Health. These are real, established names in Australian healthcare. The fact that they are willing to be publicly associated with a pre-seed startup is a strong signal.

The customer story features a video and a quote from Dylan Campher, CEO of Partnered Health. He says, "Avoca has delivered a better experience for our clients. They now have a seamless pathway into booking. And for our staff, it means they can focus on the things that really matter. With Avoca, it just makes business easy." This is a glowing testimonial, but it is also vague on specifics. The more concrete claim is the 30 bookings in half a day, which is a useful data point.

However, the evidence is still thin. The site does not disclose how many sites are using Avoca, how long they have been using it, or what the failure modes are. The 86% call resolution rate is impressive, but it is a single number. For a pre-seed company, this is understandable, but it means that the product's real-world performance is still unproven at scale.

Avoca's approach is promising because it is focused on a specific, painful problem in a specific industry. It is not a generic AI scheduling tool; it is an operational layer that learns the quirks of a radiology network and automates the patient journey from referral to results. The integration with existing RIS systems and the focus on clinic-specific intelligence are the right moves. The early customer evidence, while limited, suggests that the product is delivering real value.

The biggest question is whether the AI can maintain its performance as it scales to more sites and more complex cases. But for now, Avoca is a startup worth watching, not because it has a flashy AI demo, but because it is solving a problem that every radiology network knows well: the phone is ringing, and someone needs to answer it.