Why Voze Chose Cycle for Their Next Generation AI Platform

Voze is a sales acceleration platform designed to streamline the sales process for field sales teams. Their platform centers on capturing critical sales information efficiently, enabling reps to log essential details from their interactions in less than a minute.

When it came time to modernize the platform, the Voze team tapped industry veteran Dusty Chadwick. With his eyes set on incorporating AI in meaningful and relevant ways, Chadwick leaned on his 20+ years of experience in software engineering and management to roll out a master plan.

Along the way, he found that consuming existing IaaS services was slow and sometimes painful. This left Chadwick feeling like his team wasn't moving fast enough.

"In the fast-paced world of startups, speed is crucial. Cycle offered us the agility we needed without sacrificing performance or flexibility." - Dusty Chadwick, VP of Engineering and Head of AI @ Voze

Chadwick knew that velocity was crucial. They didn't have time to wrestle with clunky AWS services that could take weeks to set up, nor did they want to build an architecture that would become more fragile as the products grew. On top of that, they needed a solution that would keep DevOps work as efficient and optimized as possible.

"We had a lot of products to build and some really progressive applications of AI, to say I wanted my team to be fully focused on the core product is an understatement… it was my top priority", says Chadwick

With those goals and priorities in mind, let's take a look at their path that led to adopting Cycle and the stellar outcome that followed.

Transcription, LLM's, AWS, & Cycle

Traditionally, Voze had relied on up to 80 different independent contractors to handle transcription services. This wasn't just a basic speech to text translation, Voze provided each client the ability to describe their data entry processes in a very detailed way. These processes were then implemented by one of the skilled transcriptionists that would have to reference these books of instructions for how to work with each client's data. The data they were transcribing was then generally pushed on to external CRM services like Hubspot or Salesforce.

Due to the white glove, customized nature of Voze's service, when new transcriptionists were needed it would take around 6 weeks to get them onboarded. A lot of this had to do with the technical nature of the tooling they were using which required a windows workstation with a specific version of .NET installed. As Voze started to scale, they ran into limitations with how many new customer projects they could start because of these constraints around scaling the human transcription element.

Chadwick saw an opportunity to move this process to software and with the massive rise in large language models (LLMs) and natural language processing (NLP), he knew he could fully automate and solve the scaling limitations the team had run into with human contracting.

"Automating transcription with AI was a game-changer. It allowed us to enhance accuracy, speed, and scale more rapidly with the resources we have. It even opened up new markets where not having human eyes on data is a security requirement", Chadwick explained.

Integrating with existing models, the work is now completed exclusively on one of their software platforms. The time it takes to complete these transcription workloads has moved from 12-24 hours down to 15-30 seconds for processing a voice recording.

With things moving quickly, the Voze team needed a platform that was simple to use but flexible enough to not cripple their expressive, bleeding edge innovation in the AI space.

"We ran into issues on AWS very quickly, and when you're trying to do something on AWS you're almost always arm wrestling… you really have to put some time in to win. In the startup world, we don't have that extra time. Every minute counts. So I needed something that would lay on top of AWS compute, not slow us down, and be flexible enough to handle all the different unique workloads and architectures we had planned", said Chadwick.

One of Chadwicks engineers introduced him to Cycle as an option. "The way I remember it", says Dusty, "was when one of the engineers on my team knew about Cycle and had told me that it would allow us to overlay AWS, that we'd be able to get things done faster". He continued saying, "I didn't really even know Cycle at the time, I'd been exposed to using Rancher for Kubernetes in a previous DevOps role and initially the two sounded similar so I figured I'll take a demo call and let's see what these folks have".

After running through the demo, Chadwick thought that Cycle might be able to solve the top problems his team was facing:

1. Increase velocity across the board.
2. They wanted to move everything to containers to enhance scalability and portability.
3. Gain elasticity without making big engineering investments in AWS.

Voze opt'd into a proof of concept period to prove things out. This would give Chadwicks team time to deploy a few of their services and make sure it was a great fit. During that first six weeks, Voze engineers successfully migrated and brought several services from AWS to Cycle.

"I knew Cycle was the right fit for us when things that had traditionally taken us weeks started taking hours. Now we are consistently saving 10 - 20 man hours per 1 week sprint", claims Chadwick.

A Path Forward Together

After containerizing and creating the next generation of their services in Cycle, Chadwick got back to pushing his AI initiatives. He has said that they've achieved a "massive amount of scale" with their new AI platforms.

"With Cycle, my team and I get to stay laser focused on developing the next generation of technology and I think conservatively we're at least a year ahead of what most of our peers are doing in the AI space. Part of that is due to the work I don't have to worry about, thanks to Cycle", said Chadwick.

The leap in scalability has empowered Voze to handle 32x their current volume with no impacts to processing speed or scaling servers horizontally. Using Cycle allowed the engineering team to focus on pushing core features instead of fighting with ops and AWS. They also knocked out all of the bottlenecks of using a slew of different AWS services and now have a solid, containerized posture that can be iterated on quickly and effectively.

We can't wait to see what Chadwick and team will cook up next. They've been pushing the platform in unique and creative ways. If you're in a similar situation and want to see if you can have a wonderful outcome like Voze, we'd love to hear from you!

💡 Interested in trying the Cycle platform? Create your account today! Want to drop in and have a chat with the Cycle team? We'd love to have you join our public Cycle Slack community!