LISTEN TO AUDIO VERSION:
We talked with Iman Pouya, founder and Software Engineer at Voxo, about the start-up. Focus of the conversation was the business model behind the voice conversation content analysis platform, the product’s value added, in terms of the customers’ perspective, and further steps the fintech is planning on taking.
What are your backgrounds as founders and where did the idea for the business come from?
The company currently has two active founders.
I myself have a 12 year background in computer engineering both from industry and academia where I conducted my PhD. I had a great time and my professor afforded me a lot of creative freedom. I got to learn a great deal about high performance computing and distributed computing as well as using machine learning methods on a large scale. The PhD was unfortunately put on hold in the final stages as Voxo required my full attention.
Omid who holds the position as the acting COO has more than ten years experience working within marketing and business development. Before co founding Voxo, he worked in an industrial group where he led the IPO of four of the group’s tech companies.
We recently appointed Johan Wadenholt as the new CEO of Voxo. Johan was headhunted from CGI where he held the position of Director for financial services operations. Johan’s background is within computer engineering as well as industrial management at KTH and he has previous experience from another tech start-up focusing on new streaming technologies.
How would you describe Voxo in a few sentences? How do you differ from other start-ups? What is your USP?
Voxo’s service unlocks a whole new data source for our clients. Using Voxo they can now search through their voice conversations and get an overview of what is being said as well as gain insights into customer satisfaction. As we analyze each individual call, the Voxo platform can be used as a tool for an advisor or customer service manager as well as for managers who want insights from a thousand calls.
Why did you decide to specialize in ‘voice conversation content’ analysis? How do you know if there is any market potential (numbers, data and facts)?
Our methodology is highly client driven and the problem that Voxo solves originates from financial advisors who struggled with efficiency while maintaining compliance.
During our collaborations with both Nordea and BNP Paribas we tested out different solutions and validated the idea with end users. During this time we understood the commercial potential in a way that we could not have done by simply looking reports and trends. We quickly realized the value add to end users and also validated the scalability as the amount of calls and meetings was in the order of millions.
The timing for our solution was right on: the industry was starting to look for voice based solutions which aligned with our strong technical competency within that space.
Still today we listen carefully to our clients so we can continue to deliver new solutions to their problems.
In terms of the customers’ perspective, what is your product’s value added? Who have you managed to convince or win as customers?
Financial advisors can use it in their meetings to fully eliminate note taking. They will reduce admin time while increasing their compliance.
Call centers can use the system to get an overview of what is being talked about, what clients ask for and which competitors are mentioned.
Besides Nordea and BNP Paribas we work with medium sized banks in the Nordics as well as clients from other sectors and we are now looking to expand to Germany, UK, Spain and Russia.
What does your business model look like? What, when and how often do your customers pay for your services?
At the moment Voxo provides three solutions for our clients, Voxo Advisory, Voxo Meetings and Voxo Brain. Aside from these, Voxo also has a conversational chatbot in development (Voxo VICE).
Our business models differ for these services. Voxo Meetings and Voxo Advisory, for instance, are charged monthly per user while Brain is charged by the minute. It’s often difficult, but very important, to find the best model for each service. While finding the right model both minimizes costs and risks for us, it should also enable the client to scale their business together with us in an efficient way.
Subscribe to our newsletter
Get informed by our experts about what is important – every 2 weeks directly into your inbox.
Do you currently have any well-known investors? How do you finance the business?
We closed a financing round of €600,000 from NFT Ventures in March 2018, which secured our business for the closest future while moving forward. We are growing our client base and also make sure that we have healthy pricing for the projects and pilots we perform with our partners.
What technology do you use to run your product/service? Do you use your own software or do you combine various proprietary systems (AWS, AI Libraries, platforms)?
For infrastructure and generic web or mobile development, we use commoditized technology such as AWS, JAVA, python, CUDA and a lot of different open source libraries.
For the core of the platform, audio and text analysis, we mostly use technology we have developed ourselves or started off with an open source library that we rewrote completely. A lot of the existing open source tools work very well for a few of the larger languages and on a small dataset. However, our own platform is optimized to handle a large scale of audio data for many languages.
What about data privacy? Where do you obtain data from to develop your software? And how do you make sure that your customers’ data is safe?
At Voxo, we have a strong focus and great respect for data privacy and we’ve worked with several banks since before GDPR which has taught us to set up robust processes to ensure data integrity. We also offer our clients the option of on-premises installation, which enables the client to keep all information within the confines of their own data centers and firewalls. Furthermore, we have developed our own anonymization engine which gives our clients the possibility to anonymize data in real time, if needed. We have many sources for acquiring data to train our models and maintaining this diversity often requires creativity and thinking outside the box.
Finding relevant data in a large enough quantity is hard work but essential to create spot on models with high accuracy.
What further steps are you planning on taking? What are your plans, in terms of markets, customer groups, industries, events and incubators? How are you willing to develop internally, in terms of locations, the team and skill sets?
We will continue to focus on financial services as that segment is a frontrunner in AI systems investments. Since our initial collaboration with Nordea, we have been able to confirm that the FS industry has a tremendous need for new technologies such as AI, process automation and machine learning. Although this is true, we have found that other verticals and markets have started to catch up and we have acquired a few clients outside of the FS area. Starting off in Sweden we quickly received requests from all over Europe. It is also in our strategy to direct an effort towards Spain as we have identified a need for our services there. We will also continue to sell internationally from our Stockholm office.
We have a core team at the moment with strong technical and commercial execution power. During this phase there have been many hybrid roles in the company. As we grow, the organization will move to dedicated roles which will be an exciting challenge and an important change within the company.
Where do you see yourself as well as your technology in 5 years?
AI and machine learning is really starting to infiltrate most of today’s industries, creating more and more efficient processes and solutions. Our vision is that Voxo, as a front-runner within this technology, will be the natural choice for companies when selecting a provider for solutions based on AI and machine learning.
Further information on Voxo can be found on the Fintech Hub by zeb.