Martin has built his AI startup within the Microsoft ecosystem. He shares interesting perspectives on how to leverage Microsoft.
Sramana Mitra: Let’s start by having you introduce yourself as well ICS.
Martin Neale: I’m the CEO of ICS AI.
Sramana Mitra: What path have you traveled before coming here?
Martin Neale: I am a seasoned campaigner both in bringing many new technologies into the market and commercializing them by mostly working directly with Microsoft. We’re doing the same kind of thing but in the artificial intelligence arena.
Sramana Mitra: Where do you work out of?
Martin Neale: We are outside of London, close to Redding. Our workforce is largely in and around UK. Since COVID, we have become almost entirely a remote workforce although we have our headquarters building.
Sramana Mitra: What is ICS? What problem do you solve?
Martin Neale: We focus on conversational AI for the public sector. This is the public sector as we define it in the UK. That varies from geography to geography. In the UK, the public sector includes healthcare and education and the more traditional government areas that you would expect.
Sramana Mitra: Is most of your customer base in the UK?
Martin Neale: It is currently. We’re doing a little bit of work outside of the UK as well. We plan to open a US office in the next month in order to expand out. We already do some work in Germany, but our next physical presence will be in the US.
Sramana Mitra: Double-click down for me one level further and explain what kind of conversational AI solutions and use cases are you working on?
Martin Neale: The reason we focus on the public sector is, we feel that the public sector is the largest consumer of conversational AI because they have a requirement to communicate and engage at a population level. They’re under great pressure to do that digitally. It is only conversational AI that is capable of doing that. It takes on different forms. In universities, conversational AI will help around 10,000 new students a year select dorm.
We have an AI assistant give assistance to people with mental health. For the ICO, we are responsible for GDPR in the UK. We have about 350,000 people per year who are using the platform to understand whether they comply with GDPR or not.
Sramana Mitra: Where did you start? How did you build your data model? What use case did you start with?
Martin Neale: We identified the training data and the underlying AI model as the most critical factor in what we offer. What we are offering essentially is human parity AI. To do that, you’ve got to have excellent training data. We work by sector.
The first sector that we started with was for local government. We worked with a local government client who paid us for the chat platform. We built the first model for them. Our first offer was, “If you let us have access to the data, we’ll convert that to an AI model without cost.” We spent a year doing that and ended up with our first version.
That model now has over 900 different intents. These are basically different things you might do with the local government. It understands any of these various things and is able to provide answers. We use that model in all other sectors. We are often building the language models ourselves in target areas.