I am excited to share a new project with you that several business partners and I have been working on since last fall. This venture brings together our many years of experience in telecom, technology, healthcare, and call center management to solve an old problem: that it’s incredibly frustrating for patients when they can’t get through to their doctor’s office and that it’s very difficult for practices to match their call answering resources to the changing call volumes and answer all of the phone calls that come in.
We believe that patients shouldn’t be bound to 8-5 office hours to call their physician practice or wait on hold when they do call. Practices should be able to seamlessly handle patient communication, no matter the call volume.
We also believe that healthcare costs are far too high. We want to be a part of the solution (even if it’s small) to help reduce healthcare expenses by lowering practices’ administrative overhead.
That’s why we’re solving an age-old problem with a new approach – with artificial intelligence and natural language processing (NLP).
Transform9 has built the first specialty-specific, automated, conversational voice and chat bot virtual assistant for physician practices that lets patients communicate how and when they want. We’re using the same conversational engine as Amazon Alexa so patients can have natural, empathetic conversations with the voice bot, which handles instantaneous surges in call volume, books appointments accurately, and answers questions or will transfer the patient to the right person who can.
We believe Transform9 is going to transform the healthcare call center.
Our concept: What we’re trying to solve
During our time at Sequence, we saw first-hand the problem that many specialty practices have with their back office – that it’s very difficult to handle the volume of calls that come in after the weekend or after a holiday. Practices have a hard time managing the staffing levels to meet the day-to-day changes in call demand. For orthopedic practices especially, call volume usually surges on Mondays and Tuesdays because many injuries happen over the weekend. During the rest of the week, the call volume tapers off.
Our time at Sequence opened our eyes to the fact that many specialty practices are committed to answering the phones at a significant cost and are sometimes paying to have third-party vendors manage their call centers. These practices are incurring a significant financial burden to carry staff to answer the patients’ calls – not to mention all the added indirect costs of hiring, training, unforeseen sick time, personnel turnover and meeting quality standards. We know, from our time at Complete Health, the pressure physicians are under by accepting the fee schedules offered by Medicare and large insurance carriers.
Prior to Sequence, several of Transform9’s business partners and I worked together for many years at Momentum Telecom, which gave us extensive experience in telecommunications and the foundation we needed to build and operate the technology behind our concept.
When we hatched the idea for this new virtual approach, the level of artificial intelligence that we needed didn’t exist. (Alexa was just introduced in 2014. Isn’t that hard to believe that voice recognition technology has only been in our lives for a few years?)
So we waited.
By 2018, AI and NLP technology had finally matured enough to handle the majority of calls that would come in to a specialty practice, so we decided to dive in. We chose Amazon as the platform for our bot because its natural language processing (NLP) technology works well for the telephone, which requires 8 kHz for speech recognition. It also has the fastest data handoff, so there is very little latency or delay in the conversation, and it can scale to handle instant volume surges, which is important to any business. Of the options we tested, Amazon was head and shoulders above competitors for conversational interaction.
We decided to start small and target a niche that we knew very well to prove our concept, so we chose to focus on the orthopedics specialty.
Our approach: Creating virtual conversations within a telecom framework
We believe the most important part of designing and building out our business is making sure it runs like a telecom carrier (which, thankfully, we know how to do really well), as opposed to the way a software vendor might run a voice system.
We know that the phone system can’t go down. We also know what to do and how to communicate when things do go wrong, because they inevitably do. We know how to implement tight controls to deliver the right level of monitoring, alarms, security, and communication to our clients.
At Momentum, we grew the business from the ground up – at first as a telecom reseller, building software to integrate our provisioning and billing systems. Later, we implemented our own Voice over IP (VoIP) switches with BroadSoft and Sonus, which delivered call features and call routing. We used best-in-class hardware from Acme and Cisco, which supported security and networking. We built and maintained a colocation site in Atlanta and added geo-redundancy with another colocation site in Las Vegas. In the mid-2000s, we ventured into voice technology and started offering wholesale VoIP. We developed a cable modem provisioning and monitoring software and Tier 2 support used by cable operators nationwide, which moved us into the software business. Between the telephone and software services, we learned how to manage new releases and communicate with clients about updates using ITIL processes and framework. In the late 2000s, we built a business-grade VoIP offering with advanced communication features like chat and a client portal to manage the product.
All that to say, we know how to run a telecom system and a software company.
During the course of our project to build Transform9, we’ve learned a significant amount about the new technology involved in this emerging field. Over the past nine months, we’ve cultivated best practices for conversational design, which is how we script and design the bot to have natural, free-form conversations. Conversational design is changing quickly, and it’s already evolved in the short time we’ve been at it.
Using Amazon Connect and Amazon Lex to power our bot has significant benefits when it comes to machine learning models. But we’ve also run into some limitations of the technology that have allowed us to see that there are limitless opportunities to improve conversations through add-on technology, which is what we’ve done.
For Transform9, we’ve designed an architecture that overcomes some of the hurdles we’ve encountered – like, for example, that Amazon Lex is not yet HIPAA compliant. But we’ve been able to use different components for PHI data to build a HIPAA-compliant model.
Fortunately, as a small business with a boutique focus, we are able to adapt and react quickly to our clients, but also bring the experience of many years running large organizations.
The Transform9 team has leveraged their expertise in telecom, software, healthcare, and communications, among others, to bring our virtual assistant to life.
Right out of the gate, we hired a conversational designer with experience in healthcare and technology communications who could help us craft dialogue flows that feel natural, but specific to healthcare practices.
Our software development team has brought the domain knowledge of provisioning systems that has helped us design the system architecture, integrate multiple systems, and build our bot “ecosystem” that will allow us to scale, but also configure to each practice.
Recently, we brought speech pathologists on board to help us make the bot’s voice even more empathetic and natural, and now have a quality assurance team who has domain knowledge in the call center space to test the bot’s functionality.
Bringing Transform9 together
One of the biggest challenges healthcare companies face is trying to change the behavior of consumers. Many practices and healthcare networks have tried unsuccessfully to get their patients to use their healthcare portals, but in reality, people want to do what’s familiar. When they want to make an appointment, overwhelmingly, they still want to pick up the phone and talk to someone.
With our Transform9 model, we sought to make sure patients wouldn’t have to change their current behavior. Using the telephone is the historical and comfortable way to make appointments, and that’s not going to change for a very long time. Ultimately, that will change as more products include built-in smart speakers, and consumers get used to talking to bots like Google Home, Amazon Alexa, and Siri in their homes and cars. We anticipate a very long tail in terms of declining phone use, and we expect to offer an Amazon skill in the future.
With our team’s combined domain knowledge of telecom, system integration, and specialty-specific patient telephone conversations, we truly believe Transform9 is going to improve patient experiences and make it easier for them to access the healthcare they need.