ATTENTION RADIOLOGISTS: AI Is Not Going To Replace You Anytime Soon

I read this article in IoT World about connected health trends to keep an eye on for 2019. It talks about use cases for AI such as wearables, patient dashboards, surgical robots. It talks about some VS guy who a few years ago said that AI can do 80% of what doctors can do and that radiologists would be obsolete by 2022!

The overall progress of AI adoption in healthcare has been really slow. It seems to me that so many companies are trying to develop AI solutions for applications and use cases that are just too complicated or rely on an ecosystem of other technologies to make them happen.

We need to stick to the basics. No tech project or initiative will be successful unless there’s a business case to drive it. And in healthcare, the obvious business case for AI starts where IoT did: service and support.

Companies trying to deploy an IoT technology so they can remotely support their devices in the field are certainly still facing challenges today. More and more of their end users, particularly overseas, are resisting their attempts to connect to their devices on their network because of security concerns. But the business case for remote service and support still is strong.  Reducing downtime, reducing 2nd service calls, improving MTTR and FTF rates still translate right to the bottom line for OEM’s and their customers.

AI for healthcare should start here. There are AI technologies available today that can help service and support organizations improve their support capabilities even further. Some of the benefits that you can get from AI today:

-Upskilling the service organization by making your service rookie as smart as your veteran;

-Automated dispatching so that experienced techs don’t have to spend time on the phone;

-Enabling connectivity to devices without relying on the customer’s network

-Dramatically reduce 2nd service calls

 

There are many others.  And if you’re a radiologist, don’t worry – your job is safe!

How to Get Organized to Evaluate RevTwo AI

So you’re thinking about trying out RevTwo’s AI technology? We are currently working with several potential customers on proving out our technology for them and would love to work with you.

IDENTIFY KEY ISSUES

To get started, talk to your service and support teams and identify 3-5 common service/support use cases that usually result in having to send a technician onsite to resolve it.

Sometimes our customers may struggle to come up with the right issues they want to address.  One way to get your team thinking is to consider issues with one of these characteristics:

-An issue that happens often

-Something that was expensive when it happened

-Something that everyone in the organization would know about

It also doesn’t have to be something cut and dried like “the water pump failed”. Most complex issues are not that straight forward and usually have multiple root causes and symptoms. For example, it could be “my system is running slowly”.

HOW DID IT IMPACT THE ORGANIZATION?

Here it would be great if you could identify the cost of this issue, both to your organization and to your customer. It will be important to be able to show real business value if RevTwo’s AI could help you avoid or at least drastically reduce the expense of dealing with these issues. As I’ve written before in other posts, the technology doesn’t matter if you’re not solving a business problem.

One example would be to use reduced resolution time as a business case. In that instance, you look at each issue and calculate its resolution cost:  Cost = resolution time * labor rate. Then in the POC we would observe a much lower resolution time which would translate to cost savings.

WHAT IS THE ROOT CAUSE?

Ultimately, what was the root cause? Some use cases have many root causes, making them more complicated to diagnose, some do not.

HOW WAS IT DIAGNOSED?

How was the root cause identified? What things were asked about, what was looked at, tested, etc?

What would also be extremely helpful is to list out the diagnostic indicators that were used.  Examples would be sensor readings, log files, registry settings, and where they are located (ie file, table, etc).

HOW WAS IT FIXED?

What steps were taken to fix the issue? If you have a knowledge base, where is the article that explains the solution? View only access to your KB would be helpful as well.

WHO IS THE EXPERT?

All organizations have that one person (or many of them!) who has been around for a long time and has seen it all. Ultimately, you want the RevTwo AI to be as smart and knowledgeable as that person so that the entire service and support team can have access to that know – how.  We often spend some of our time talking with that person in order to make sure that the best solution set gets input into our AI database. So access to this person would be important.

 

If you want to learn more about how RevTwo AI can help your service and support organization reduce triage and resolution time, email me at dave@revtwo.com

How RevTwo Acquires Data: Overview

Overview

Complex products have complex problems!

Unfortunately, the only way to solve such problems today is with highly trained personnel and time. Current methods include:

1) On product data and event logging
2) Remote product access
3) Remote field tech support
4) IoT cloud connectivity

Today’s Challenges

All with varying success and their own unique set of challenges. For example:

Product logs are useful but finding the right information in the right log can often be time-consuming and frustrating.

Remote Access, which provides skilled technicians with access to physical products, is becoming less and less available as the cybersecurity threats of today’s world multiply and the risks associated with a data breach grow. Even with access, technicians still face the challenge of identifying the right diagnostic information and using their expertise to parse its meaning.

Remote Field Technician support provides a more experienced product expert at the disposal of field technicians who have varying skill and training level, to aid in the troubleshooting process while onsite. While this helps to prevent site revisits, it increases the costs of already expensive support activity.

Continue reading “How RevTwo Acquires Data: Overview”

How RevTwo Acquires Data: Walk-Up

Walk-up data acquisition is built upon two pillars:

1) Physical proximity to the product, provided by the customer themselves or a field service technician.

2) Mobile invocation. A mobile support app is a very convenient vehicle for accessing data. It is easy to distribute, provides a high degree of security, and is easy to use. RevTwo uses mobile apps to provide walk-up data acquisition.

Walk-up data acquisition is ideal in situations where the physical product does not have access to the Internet, because it is in a secure environment, or because Internet access is not built into the product.

In a walk-up scenario, product data can be obtained in one of three ways.

Ad-Hoc Networking

Ad-hoc data access works by forming a transient connection to the product itself. The underlying communication technology for these transient connections is usually Wi-Fi or Bluetooth, but other mechanisms could be developed. The general process is to:

1) Identify the connection available for a particular product
2) Form a secure connection to the product
3) Request diagnostic information via that secure connection
4) Obtain that diagnostic information
5) Shut down the connection
6) Use the information to perform an AI analysis

Continue reading “How RevTwo Acquires Data: Walk-Up”

How RevTwo Acquires Data: In-Product

In the in-product scenario, RevTwo acquires data from the product itself. While this sounds similar to most IoT implementations, there are some pretty significant differences that lower the barrier to deployment acceptance. First, RevTwo is provided in a customer-brandable helper application. The helper application has a user interface that manages the support workflow.

Unless requested by the customer, RevTwo is inert and does not communicate. No data is posted, remote access is not possible.

When the customer has a support issue, the RevTwo Support Application is typically invoked via a link on the product’s own user interface or via the operating system tray.

Support Helper Application

The RevTwo Support Helper Application contains many standard features, including built-in

• Data collection
• AI invocation
• Chat
• Optional customer controlled remote connectivity
• Optional customer-controlled VoIP support

For the purpose of this discussion, we are going to focus on the workflow that happens when a customer requests support. In this scenario, the Support Helper Application does the following actions:

Continue reading “How RevTwo Acquires Data: In-Product”

How RevTwo Acquires Data: Cloud

The next data acquisition mechanism supported by RevTwo is via a cloud data source. Cloud data access can be optionally performed for each issue submitted to RevTwo. Each product type has its own cloud data collection mechanism and is invoked regardless of the presence of an existing data set. In fact, the cloud data acquisition has the ability to modify the data set submitted with the issue before AI evaluation. Each issue that utilizes cloud data acquisition, goes into a pending state until the cloud data request is satisfied.

The RevTwo cloud data access mechanism is not limited to a single data source.

 

Continue reading “How RevTwo Acquires Data: Cloud”

How RevTwo Acquires Data: API

This final data acquisition mechanism is simply to attach the data to a RevTwo AI request. RevTwo allows API interaction with the AI directly. The API allows the attachment of a data dictionary and can be invoked from any requesting source.

How is R2 AI different from other AI’s?

When we talk to customers, we are often asked how we are different from other AI solutions out there.

AI solutions have been around for a while. If you’ve ever gone to a website and needed support for a product that you purchased, when you select the chat feature, you’re talking to a chatbot, not a human. A chatbot employs AI that takes the text you’ve typed in and matches it to a knowledge base and then asks more questions until it gives you an answer. This use case is mostly used in B2C businesses.

Continue reading “How is R2 AI different from other AI’s?”

AI Isn’t as Hard to Implement as People Think

AI isn’t hard.  It can help you today.

I am reading more and more about how Artificial Intelligence (AI for those of you who have not been online in awhile) is going to change the world.  And soon, like the end of next year, according to this article.

But in my opinion, when people think of things like AI, they think of things like chatbots, or in-app support.  Indeed, these tools can perform many useful functions that humans once had to do.  But chatbots can only help with simple issues.  If the issue is complex, then humans have to get involved.

Continue reading “AI Isn’t as Hard to Implement as People Think”