Maeve Bleahene, co-founder and Head of Operations at Peroptyx
I spent the previous 15 years running a global recruitment function for a multinational business that spanned 110 countries. It was quite a substantial operation and that is where I really came to understand the importance of having the right people, the right culture, the right contracting model, and technology for the business.
It also highlighted the importance of finding the right way to engage and ensure that customers were consistently satisfied with the services and solutions we were designing and delivering.
Getting involved in Tech is still a hot topic for women, what was the motivation for you?
It’s a great question - I have just finished a podcast titled ‘Equilibrium in AI’ where I chat about the challenges with the lack of female representation in Machine Learning and AI – it currently runs at about 15 percent. The issue here is that the representation itself has an impact in terms of the rules and structures in terms of how AI systems are developed and that today as are inherently, and unavoidably, male-biased. I also feel that AI/ML tech ought to be more accessible to women – for example through specifically targeted outreach initiatives – to create downstream opportunities for leadership roles to bring about a more gender-balanced sector. For example, 3 out of the last 4 senior hires at Peroptyx have been female and even with that, we expect to achieve a 50/50 balance by year’s end.
In terms of my own involvement in tech - having worked with many leading global technology brands in my last role over 15 years, I experienced a shift in focus towards data in the mid-2000’s and developed a deeper understanding of its importance in AI systems. My question was “OK, I get that good data is essential to building good AI systems, but who is going to evaluate whether or not these AI systems work in real-world situations involving, you know, humans?!”
This question continued to drive my thinking around how this could be done in an ethical, secure, open but innovative way, which led me and my co-founder, Paul, to quit our jobs and start Peroptyx. We assembled a team and have spent a couple of years building the global infrastructure required to offer medium and large global enterprises an independent, real-world assessment of their ML and AI driven platforms and services at a global scale.
How have you found your journey so far and what has been your greatest accomplishment?
The journey has been exciting and unpredictable! Exciting because we have developed an innovative solution that helps global consumer technology brands deliver world-class performance for their ML-driven platforms and services in any market or markets. Unpredictable because many of the smaller enterprises are still struggling to get their ML models into production and need help solving the data and data quality equation. So, we make the lives of product owners and heads of data science much less stressful by sourcing, qualifying and retaining in-market domain-experts as if they were part of the Customer’s release team, to prepare or enhance their data and then evaluate the relevance of their model outputs integration into their platforms, services and content. The crucial element and our USP at Peroptyx is our ability to retain the most relevant domain experts over time, so customers experience a competitive, easy to consume yet highly customised on-demand offering.
Our expertise in quickly and accurately identifying and addressing bias in data and models that power AI drives us on our mission - to help our customers improve the technology experience for humans across cultures and countries
I am most proud of the team I have helped put together – if I may, we have a decent pedigree having worked for the leading consumer-facing enterprise technology companies at various stages of our careers. We have all learned from past mistakes and are passionate about using that experience to share our vision, to attract people interested in developing the most innovative ways of solving data-centric customer problems as part of their everyday experience at Peroptyx.
What has been the most challenging thing so far?
Convincing our enterprise partners to build enterprise infrastructure at start-up prices!
You have created a business that is adaptable and progressive. What has that process been like for you?
Hard work and discipline are required to validate the value we deliver for customers a global scale. The challenge was building a model that could adapt to the nuances of the legal, contractual, financial and tax regulations in the markets that we are operating in today and those that our future customers expect us to be in tomorrow.
For example, new and evolving data privacy legislation must be considered and complied with. As a new business, we were able to develop a data privacy-driven security model from the ground up, which was essential to developing a unique value proposition to address customer pain points around global data privacy compliance regulations. Our company culture has been built around privacy values, and customer benefit from working with a team that’s laser-focused on preserving and protecting the integrity of customer data.
There was quite a lot of testing of the business model itself. For example, in terms of its efficiency around practical implementation and scalability across global markets. Getting that right and properly de-risked before we launched was important – our platform is currently configured to manage teams in 20 markets across The Americas, EMEA and Asia Pacific, and we can add capacity in new markets in a matter of days instead of weeks and months. This matters if you are a unicorn business growing revenues at 40% a quarter and targeting consumers in 10 new markets a year with your ML driven platform or service. It also matters if you are a global business with a product or service that is underperforming, and customers are disengaging due to poor local relevance of the ML-driven outputs.
The adaptive part is around the culture you must build to implement a model like this, simply because operating in multiple markets is challenging if it’s not properly designed and thought through from the get-go. What I mean here is the right culture for a team that can understand and switch conversations around how things ought to work in each market based on the knowledge and expertise we have developed. This has been important in the adaptive part of the business.
Who is your target audience?
Our target audience are global internet platforms and enterprises turning over 50 -100 million dollars a year in revenues. They are using machine learning to expand their customer facing services and platforms to new markets and need a partner to scale and help them maintain the growth rates with a solution that delivers Internet platforms companies are our second target audience. We have designed solutions for these types of companies in the past. Today we have built a lot more substantial privacy controls, a lot more robust components in our process offering solutions that exceed customers market-specific data quality, data refresh and model evaluation requirements.
What advice would you give to fellow entrepreneurs stuck on mapping out a strong business model?
Mapping out your business model and testing it are two different things. Along with identifying your target audience, developing your value proposition, developing strong partnerships one of the things that I would advocate is to read, research and revise your model based on best practices and based on other people’s experiences. I think it was Mark Twain that said, “the man who does not read has no advantage over the man that cannot read” and so for me it’s about making sure that reading and researching the situations where you are stuck is critically important. Seeking out people who can help you with the problem is also really important. There are lots of valuable resources available for start-up businesses today like Enterprise Ireland, Western Development Commission (for businesses based in the west of Ireland) & New Frontiers Programme to mention a few.
What plans do you have for Peroptyx over the next two years?
We have spent the last 18 months building a world-class infrastructure. As I mentioned we have come from running and scaling an enterprise business in the past and we’ve applied all of the experience telling ourselves- wouldn’t it be great if we had this, or we should really automate that, or we can really innovate here with this…and that’s exactly what we have built and are really excited about demonstrating this to prospective customers. So, for the next two years, it’s about scaling it, reaching into new markets and new customers and so the sales cycle is now the key cycle for us for the next 12-18 months.
