Often, new solutions for call center staff claim that they’re powered by AI when in reality, they’re just hoping to capitalize on a buzzworthy phrase. At best, these tools may offer a simple wrapper around an existing GPT model and call it a day. 

This matters because GPT on its own cannot do what FastTrack™, the new AI copilot from Infinitus, can. Without effective guardrails and augmentation, GPT can’t handle unintuitive flows and latency constraints, and may even hallucinate. 

FastTrack enables call center and patient support staff contacting payors to bypass tedious IVR systems and hold times. This is possible because of the Infinitus AI system, which employs domain context, complex modeling, and intelligent engineering to effectively manage the many challenges involved in navigating healthcare phone calls. 

We’re diving deeper into what this means below.

IVR, meet AI 

As soon as a FastTrack user queues up a call, they’re informed of estimated payor hold times. This is something Infinitus understands thanks to our proprietary knowledge graph and experience automating millions of calls. 

Once a call begins, it likely starts with a payor’s Interactive Voice Recognition (IVR) system. IVRs are notoriously difficult by design and often created to minimize the number of callers who reach a live agent. FastTrack navigates these systems as if it were a human, following commands to “press 1 to speak to an agent,” for example. But doing so requires a blend of complex modeling and constraints around latency. 

To successfully traverse an IVR system, FastTrack must first process audio data – the IVR recording – which it converts into text with a speech-to-text model. From there, the Infinitus AI system understands the text and determines the appropriate action to take. It is able to do so because of our knowledge of task types, subject matter, situational contexts, and resulting possible actions. 

On top of this, FastTrack must understand how to extract information when unexpected situations arise. Say, for example, the payor doesn’t have data for a patient in their system, or the payor is unexpectedly closed – FastTrack must appropriately handle such edge cases.

From an AI and machine learning perspective, IVRs are extremely challenging, in great part because they’re deliberately built to not be intuitive. FastTrack must be able to understand what to say or which number to press instantly, even if that action is unintuitive – in some cases, a call will be ended if an IVR’s prompt isn’t answered within a few seconds. It is the Infinitus AI system that makes all of this possible.

Is it an ad or an agent?

Once FastTrack has completed a payor’s IVR maze, there’s a high chance the Infinitus AI copilot will then be sent to wait on hold for an agent. Though the hold phase of a call isn’t designed to be unintuitive, it’s still a challenge for AI to navigate.

The time spent on hold can include a number of different elements – music (with or without lyrics), advertisements, and updates on estimated wait times, to name a few possibilities. If FastTrack’s job were as simple as just transcribing everything it “heard” at this stage, it’s easy to imagine what a jumbled mess its output could look like. 

Instead, Infinitus employs in-house multi-modal AI models at this stage that take into account both audio and text signals to understand what is actually happening on the other end of the line. Thus, we can understand what’s an advertisement vs. what’s music vs. what is an actual agent joining the call. This is the AI magic that enables FastTrack to gracefully drop human healthcare workers into a call only once a live payor agent is present. 

Take note: The summary is AI too

FastTrack doesn’t just save front- and back-office healthcare staff time they would otherwise spend traversing IVRs and waiting on hold. It also automates the creation of call summary notes, so FastTrack users don’t have to manually write them at the conclusion of a call and can instead move on to their next task. 

These summaries are possible because our AI system understands the most important elements of the call transcript to note, and automatically includes them. This is yet another feature that wouldn’t be possible without a system purpose-built for healthcare settings.

Employees can be up-and-running on FastTrack in one week with a no-integration option, but integrations are available via API and with popular systems of record – further enhancing efficiency and ease of use. To see the AI copilot in action and see for yourself how it can save your team time, increase productivity, and even boost morale, schedule time with us today.