In How TD Bank avoided data swamps and built their digital engine with Talend, I wrote about Talend's "bold vision of modern architecture, IT cost reduction, and shifting legacy investments into something better. "
But what is that "something better?" That depends on the company, but following this vision along, a core goal is using that shifting IT budget to build out apps that matter. Eventually these apps might look more like microservices, managed via a container-based architecture, but that's a matter of IT philosophy/evolution that's hardly set it stone.
What isn't negotiable in this vision is a platform. You might start small and go for a quick win, but you want the process to be "repeatable and scalable," as the cool kids put it.
The key is that IT operational efficiencies aren't enough. There has to be a "digital engine" constructed in the process, one that delivers insights to users and apps to customers - without a multi-year, rip-and-replace
bog pit overhaul.
As I moved on from Talend Connect to Couchbase Connect last week, the continuity was striking. Little wonder given that Couchbase, a NoSQL database with bold enterprise intentions of its own, is a Talend partner. Both companies unveiled Kubernetes plans last week that they believe will put them ahead of the game in this IT evolution. For now, I was more interested in customer use cases.
One thing that bothers me about so-called "next-gen apps": they aren't ambitious enough. They tend to focus on convenience rather than fundamentally changing how we live and work.
BD Medical Technology's Briight app, designed for people managing diabetes, is exactly what the doctor ordered. Billed as "Giving patients with diabetes the freedom to be a person with diabetes," Briight is now available on the Android and Apple stores.
Briight - the lean origins of an app with a mission
The app was built in an iterative style by a digitally-focused group embedded in BD called Digital Health. At Couchbase Connect, I had a chance to talk to Bryan Memmelaar, Head of Software Development at BD Digital Health, about how he cultivates a startup-like environment inside of BD.
Digital Health was set up to "operate differently" inside of BD, and move much faster with software development:
If you look at BD holistically they have longer project plans, medical devices, and large QA cycles. We needed to be the speed of digital.
As for my apps-with-a-purpose bias, that's what drives Memmelaar:
That's honestly what brought me to BD was the mission and the impact, because it's a large company with a worldwide impact - and a good mission to have.
As he put it during his presentation:
We're on a mission to create unique and personalized digital care experiences. We want to empower the patient to take control of their health and improve their outcome.
Briight started out as "Leah," an iterative prototype used by a small group of early adopters. But that's changed now. Today Briight, now powered by Couchbase, is used by hundreds of thousands, with lots of room for growth ahead.
I am un-enthused about the fitness apps that monitor your data exhaust. A number of diabetes apps are similar. They are helpful to a point, but are ultimately data loggers. Briight tracks data and provides analytics, but it's intended to be much more than that. As Memmelaar told attendees:
BD Briight is your always-on personal diabetes companion. It's the freedom to be a person with diabetes, and not just a patient with diabetes.
For people with these conditions, life can be isolating. And let's face it, even if you're fortunate to have good medical care, you're probably not getting the personal experience of a doctor looking in on you all the time. So can technology break down some of that isolation?
You hit it on the head. We didn't want to create just another data logging app, because it's impersonal - and why am I going to use that, versus the other fifty apps that are out there?
They took a fresh look at the problems diabetics are coping with:
We really said, "Let's take a step back. How do we engage the user, and how do we show them that we're actually there to help?" For diabetics, it's a very lonely disease sometimes, so how can we allow them to have an avenue where they can ask questions that maybe they wouldn't want ask to their caregiver?
Performance issues prompt a move to Couchbase
The early version of the Briight app, dubbed "Leah" at the time, was typical of the Digital Health model: lean/agile development, rapid iterations. Leah launched a year and a half ago, which brought in feedback fast and furious. But as Leah, Memmelaar's team ran into performance barriers - and that's where Couchbase comes in:
It was a evolution. I knew we had to look at something else, to be perfectly honest with you.
Memmelaar didn't know about Couchbase at the time. He asked around internally; his R&D group had done some Couchbase work.
As I started looking at the other options for data synchronization, it just became more and more evident that Couchbase would be a good direction.
Prior to Couchbase, Memmelaar's team struggled with data synchronization with the app, including offline sync. Sometimes their rules engine fired incorrectly based on insufficient data, and the microservices needed to "chat" with each other "too much" to make sure they were operating on the same data.
After the Couchbase implementation, those problems were resolved. For Briight version 2, running on Couchbase and hosted on Azure, the benefits included:
- no single point of failure
- offline capability
- addressed the synchronization challenge and abstracted
- "took away the hard challenge of HTTP delivery"
- product roadmap alignment, including support for full text search
- supports high availability and global replication (needed for a pending global rollout beyond the U.S.)
- supports built-in auditing, role-based access control and encrypted communication
Basically, Couchbase brings all the stuff that allows the Briight app to scale, while providing the performance consumers expect. The microservices architecture remains intact, but it's now working off of a single Couchbase document store. BD Digital Health has also implemented Kubernetes for the deployment.
Machine learning drives the personalization
Briight's user adoption is growing fast. The app is now robust enough to provide an end-to-end diabetes management experience for users. That includes a machine-learning-enabled chatbot that provides personalized content for each user. Type 1 and Type 2 diabetes patients see different content, and the personalization flows from there.
The machine learning aspects are ambitious, aiming for "guided digital conversations that uncover the "why" behind a glucose reading."
Memmelaar says that after a couple of iterations, the machine learning has gotten pretty sophisticated, including a content management system. Each piece of content is tagged, with rules on tags and content prioritization. Those machine learning elements come into play during responsive chat sessions, where relevant content is served up:
The wrap - on privacy, platforms, and a recruitment pitch
When it comes to medical apps, privacy is the utmost concern - and biggest barrier to adoption. Briight is designed to overcome these concerns; the app doesn't ask for a lot of information up front. The app wants to earn your trust, providing good info that wins over users. Integration with other BD medical devices also wins trust by connecting to a proven brand. BD's global compliance team is tasked with making sure the app is compliant across local data regulations.
Not too shabby for a 20 person team amidst a company that's now 90,000 employees across fifty countries. But will Digital Health's way of working impact the larger organization? Time will tell, but:
People are starting to take notice. They are asking, "How is this team deploying every month, how are they operating differently?"
There's plenty of work ahead for Memmelaar's team. It's not just about one app, it's about the platform:
We're building client experiences on top of a common platform. The platform comes first. Scalability needs to be there, and then we'll build out experiences on top of it... Part of our decision to go with Couchbase was the ability to scale our content globally.