Xyonix

BACKGROUND

Introduce your business and what you do there.

I’m the BI team and data analytics manager for Delta Dental of Washington, a dental-only insurer. 

OPPORTUNITY / CHALLENGE

What challenge were you trying to address with Xyonix?

Our business was divided into group and individual pipelines. Our individual business posed a unique challenge in that we wanted to understand how individuals were using our services. We also wanted to track attrition, predict coverage termination, and understand the factors that affect such an event. Later on, when a new director came to our marketing department, we flipped the project and turned toward understanding the factors that convinced people to stick with their individual coverage. 

SOLUTION

What was the scope of their involvement?

Xyonix started with what they called the Pathfinder phase; they met with our stakeholders, leaders, and technical people to get a true understanding of the scope of work and our goals. After that, we gave them representative data about our pipelines. They determined whether we had the right data to move forward with our goals by analyzing the data set's type, quality, and breadth. This phase revolved around predicting coverage termination, and it lasted for two weeks. We switched to a new approach because our marketing team couldn’t demonstrate the value of their initial findings to our prospective clients. 

During the second phase of the engagement, we gave them a larger, more representative data set. They determined whether our data provided “signals” that could help us achieve our business goals. This phase lasted for a few months as their team refined the data. We also provided additional sources that would be helpful in their analysis. Eventually, we delivered their findings to our leaders, who discussed their usefulness. This was crucial in our decision to engage them on a longer-term basis to create a production-ready model. 

We’ve decided to move into an ongoing relationship with them so that we can partner up for other use cases. In our retainer contract, Xyonix is only maintaining the two data models they’ve already built. Meanwhile, our new marketing director is working with them to put together a business case that can demonstrate the value of their findings. If this is successful, we’ll go back to developing the data models.

What is the team composition?

We mainly work with Deep (Founder & CEO) and William (Senior Data Scientist). 

How did you come to work with Xyonix?

We found them among the three companies that were referred to us. At that time, a lot was happening in the machine learning (ML) and AI space. As a result, it was difficult to find a company that was willing to take us on, especially since we wanted to move quickly.

Among the three companies, Xyonix followed up quickly with us and treated our business as important. Their team was also easy to work with from the get-go. By contrast, one of the other companies never even contacted us, while the other one had a backlog that wouldn’t allow them to do anything for us in a timely manner. As a result, we chose Xyonix. 

What is the status of this engagement?

We started working with them in November 2019, and our relationship is ongoing.

RESULTS & FEEDBACK

What evidence can you share that demonstrates the impact of the engagement? 

While the first project didn’t succeed, the Xyonix team has produced stellar results for the second phase of the engagement. Our company has only dipped our toes into the world of data science and ML, so we don’t have specific success metrics other than the deliverables. So far, they’ve been able to achieve the 70% accuracy that we want to have. They also give us timely information about our progress along the way to ensure we’re not investing unwisely. 

Further, they’re truly helpful in terms of educating me and my team on how to shape data for ML. We’ve done plenty of reporting and ad hoc analysis, but data science is different, so having Xyonix on our side has been invaluable to us. In fact, in our current contract, they’re going to help us interview and hire more data scientists internally. Still, we’ll continue using Xyonix to help our resources and teams to bring new data models into production. 

How did Xyonix perform from a project management standpoint?

Their project management style is excellent. We’ve been meeting with them fairly regularly every 2–3 weeks to review our progress, answer questions, discuss findings, and make up plans on what they need to accomplish by our next meeting. In between such meetings, we have senior-level check-ins as we find information to bring forward. 

Their materials are always well-prepared and easy to refer back to; the presentations have a high level of detail that makes them highly usable beyond our status meetings. Moreover, the team members are highly proactive, and we’ve never had issues setting up impromptu meetings when questions arose between status meetings. 

Moreover, their team members are great at speaking with their audience. When talking to our technical team, they can get into the nitty-gritty of the process to help us learn. On top of that, they’re truly spot on when presenting to our senior leadership. According to one of our leaders, it’s easy to understand what Xyonix is doing and where they’re heading. 

What did you find most impressive about them?

Xyonix’s most impressive characteristic is their passion for helping us solve our problems. They truly believe in what we’re trying to achieve — they’re excited about it. Their members see the value of our work and understand our vision. 

Likewise, they want to engage us in a way that’s beneficial to our business. Their team is aware that whatever we build can be extended to our larger organization. In other words, they’re passionate about making our success accessible to others. 

Are there any areas they could improve?

We’ve had mistakes on both sides. Early on, we learned that the marketing department should’ve been more involved in validating the type of business cases in which we’d use Xyonix’s findings. We shouldn’t have waited on the results of their analysis before deciding on such business cases. Their team could have articulated this point, and our team should have understood it, too.  

Do you have any advice for potential customers?

Make sure that you have a clear vision and goal in mind. That being said, such visions can be broad, which makes it hard for you and them to focus on a more specific direction. As such, be prepared to speak about more precise goals that you want to achieve and clarify them upfront. 

BACKGROUND

Please describe your company and your position there.

I am the CTO at a startup studio called Madrona Venture Labs

OPPORTUNITY / CHALLENGE

For what projects/services did your company hire Xyonix, and what were your goals?

We had a broad product idea that required audio and video processing and natural language processing (NLP) expertise, as well as experience building cloud-based, scalable model training, and inference pipelines. We also needed a team that had product chops and could help connect the product and user experience to the underlying machine learning concepts.

SOLUTION

How did you select this vendor and what were the deciding factors?

I have known Deep for a number of years, having collaborated with him on similar vertical machine learning products. I knew he and his team had the chops to build out this critical side of the product and underlying machine learning technology.

Describe the scope of work in detail, including the project steps, key deliverables, and technologies used.

Like many machine learning-driven products, this project required iterative development, starting simple and basic, adding complexity as more data (esp labeled) is accumulated. More data and examples enabled more features and sophisticated machine learning techniques. A POC was up and going quickly, helping us test the user experience and brainstorm more product features. The pragmatic approach taken by Xyonix helped us manage risk and timelines which enabled us to get something into the user's hands for testing sooner rather than later. I believe the technologies were AWS cloud infrastructure along with open source machine learning and data science tools.

How many people from the vendor's team worked with you, and what were their positions?

Mainly 2 people from Xyonix, Deep and his associate Bill. Both have strong backgrounds in tech, machine learning and product development.

RESULTS & FEEDBACK

Can you share any measurable outcomes of the project or general feedback about the deliverables?

What was probably most important in the early stages of product development was the ability to stub in simple versions of the signals identified and extracted by the analytics and machine learning technologies. From there, the application would get better and better each week.

Describe their project management style, including communication tools and timeliness.

Deep and Bill were always well prepared for our meetings, cognizant of the delivery schedule, communicative with us when there would be slippage. They also were forthright on the probabilistic nature of developing solutions to challenging problems. A lot of the work had some experimental or R&D component to it. Deep and Bill never tried to handwave or paint rosy pictures - if they thought something was going to be hard, they were vocal about it.

What did you find most impressive or unique about this company?

They are insightful of the rapidly changing and evolving space of machine learning technology and solutions. Importantly, they are also pragmatic as well as affordable. Their hands-on, practical experience makes them a great choice for startups who are trying to get up and off the ground quickly and helping find product-market-fit. They are also fair, reasonable, and good human beings.

Are there any areas for improvement or something they could have done differently?

Nothing comes to mind.