TechRound was delighted to speak the founders of Filmily, Ian Kenny and Andy Doyle, about their promising startup. Their company, created in 2017 is a fan-based platform that automates the process of editing content and turns it into engaging films. How do they achieve all this? Through machine learning.
What was the main inspiration behind Filmily?
Filmly started initially with a friend asking me if there were any good apps to use for collecting pictures and videos from wedding guests. After researching there were a few apps, but they didn’t seem great and didn’t create films. They were more focused on just presenting what was recorded. The idea of understanding that we can learn what was recorded and do something better evolved after looking at various Computer Vision and AI services available by Silicon Valley giants. In a conversation with Ian, he instantly understood the bigger, more commercial side of Filmily, for large events around sport and music. This changed an idea into a business.
What are your work backgrounds and what were you doing prior to starting Filmily?
I was a software developer for many years, in product development and working with TV companies. We spent a lot of time understanding video, how it can be streamed, managed, etc. This was a great basis for Filmily. For the last 12 years, I have run my own business, acquired companies, and grown a team, who are the team building the platform. Ian has a background of over 25 years in the commercial side of advertising and media, initially working for global agencies with well-known brands. When I met Ian in 2003 he was heading up commercial aspects of a consultancy business around TV. Ian founded his own business, and gained some global clients, and sold the business.
What was the process of starting a business like, any surprising challenges that have come up?
As we have both founded businesses separately, we knew the trials and pitfalls of starting a company
. What was very different in this case was that we were not starting an agency, or service-based project business, but a product-focused company. For both of us, it was a new experience of on boarding investors and understanding a business value at different stages. We were very lucky to have received an Innovate UK grant to kick this off.
What makes Filmily different from its competitors?
We are the only platform that has an automated solution for creating content for multiple mobile phones. We know we are the only platform as we have a Patent published to that effect! There are other platforms out there, but they use a lot of manual processes and teams to edit and make content. We speed this up immensely. Due to the automation, we can scale much quicker than our competitors.
Is there anything that you both wish you knew before starting Filmily?
From a technical perspective, the main thing was not to not create what we call ‘technical debt’. Its very easy to just build something quickly and easily, without always considering the bigger picture, or future requirements. We are lucky with our experience we haven’t fallen into this too much, but there have been elements where we have to go back and rework.
It would have also been informative to have a much deeper understanding of the different requirements and pinch points with different types of sport. We have learnt this now from working closely with 1. FC Koln on an accelerator program in Germany with HYPE.
Ian: When you have a good idea it attracts a lot of interest and you need to be selective about the advice that you take. Identify people that can help the journey but be cautious about how you use them.
What does the future look like for Filmily, do you have plans for expansion?
We have a long pipeline of clients who we will be working with, across sports and weddings. Some of these are game changers for FIlmily as a business. The platform is in a position where we are very confident with it, and able to serve clients amazingly. This is a business
that will grow quickly once we are live across a number of vertical markets. By the end of 2020 we will be much more well known. The key for us is that day when we don’t have to explain who we are or what we do, but someone just asks for Filmily as part of their fan engagement plan…
What piece of advice would Filmily give someone looking to start a business in machine learning?
The first thing is to go and ask real clients what they need and expect from a service before writing any code. Understand what the outcome needs to be. Second, is the question, ‘how do you obtain and get training data? This is not as simple as most thing and usually forgotten. ML is only possible with a definite plan for getting consistent data to learn from.