As experts in the complex and fast-moving world of machine learning operations, or ‘MLOps’ for those in the know, the team behind Fuzzy Labs are self-confessed nerds. They come from an engineering background, experienced at putting technical systems into production and helping companies scale.
Technology is at the core of what they know and love, with Fuzzy Labs being the enabler of this passion. And, while they are of course committed to growing Fuzzy Labs, their drive goes beyond revenue growth and quarterly results, as the company’s co-founder and CEO Tom Stockton explains:
“Money’s not the primary motivator for us. We are technologists and are interested in doing interesting things with technology, but we want to do that with impact. AI/ML offered a way into this. We’re interested in what impact we can have with this new wave of technology; that’s what drove us to start the business.”
There is certainly a clear need for their expertise as AI and machine learning technologies have continued to become more commonplace. Being able to mature the use of these technologies is now a key consideration for organisations, presenting an engineering gap that requires specialist skills – a gap that Fuzzy Labs is well placed to fill.
Delivering value from AI
The focus of the team at Fuzzy Labs is firmly on making AI proof of concepts practical – helping organisations take those initial proofs or early stage prototypes all the way through to production, at scale, in a secure way.
There are various phases to accomplish and challenges to solve along that journey, with ML models differing from traditional software in several ways. For example, in addition to being software, ML models are tied to data. This introduces new additional practices that need to be addressed such as tracking data versions and changes to ensure data control.
What’s more, the process of training a model isn’t like compiling a piece of software. According to Fuzzy Labs’ co-founder and CTO, Matt Squire: “It’s more like running an experiment where you input some data and configuration, but you don’t know how well that model will perform. So you train the model, then you find out the impact – that’s your experiment. Typically you’re going to run lots of those experiments, so you want to be able to keep track of them.”
“We’ve got to think about a data scientist’s daily activities – whether that’s analysing data, tracking experiments or monitoring performance – and work out how MLOps can help them do that more effectively.”Matt Squire, co-founder and CTO, Fuzzy Labs
When in production, the challenge then is monitoring the model to ensure that it’s doing the right thing and understand how to scale it. These are all areas in which Fuzzy Labs gets heavily involved for its customers, leveraging its MLOps expertise to productionalise machine learning, enable data scientists to work in more effective ways, and help bring about the cultural changes required.
As Tom explains, this is all in pursuit of “getting ML to a point where it can deliver value. Otherwise you end up with experiments that never see the light of day, or deploying something that hasn’t got any stability around it.”
To achieve this for customers, Fuzzy Labs differentiates itself from other MLOps organisations in two key areas. First is its focus on open source technologies. The company has put a lot of time into researching the MLOps tooling space and now uses its expertise of those open source tools to help customers operationalise their use of AI.
The other area, which Matt highlights, is its commitment to being innovation-led: “Innovation is something that we’ve actively investing in throughout our journey. We put dedicated people onto innovation projects and look to push the state of the art in MLOps – what we learn from innovation can inform how we help a customer to deploy, scale, or secure a model. We’re looking not just to be implementers of MLOps, but to be creating new standards as we go along.”
This commitment to innovation is demonstrated in the company’s new initiative Fuzzy Labs Squared – a vehicle for innovation, research, and collaboration across industry and academia. The goal is to research novel areas of AI/ML, taking ideas from partners and customers to solve real-world challenges in areas such as health and the environment.
The power of collaboration
A specific domain that is quickly becoming a more important pillar of Fuzzy Labs’ activities is secure government. This is a notoriously difficult sector to break into given the sensitivities involved, which is one of the reasons for Fuzzy Labs choosing to work with BAE Systems Digital Intelligence.
For any small consultancy, industry relationships offer a way to scale faster than they could organically. But perhaps even more importantly when considering a complex and restricted industry like secure government, the right collaborator can provide invaluable insights and reach when entering the market.
“We’re quite vertical-agnostic rather than being an expert in specific verticals,” Tom explains. “We have deep technical expertise, but our customers are the domain experts. So, when we’re going into a new strategic vertical, collaborations give us access to specialised knowledge that we can tap into to boost our understanding of the nuances of the market and how to work with these customers.”
“BAE Systems has enabled us to get our message to the relevant people. It’s like a stamp of approval that provides assurance to customers.”Tom Stockton co-founder and CEO, Fuzzy Labs
Credibility is another important factor. Along with knowing who to speak to and how to speak to them, secure government customers need to be able to trust what they’re hearing – a perennial challenge for small businesses.
As Tom describes, BAE Systems has provided considerable value in this area: “Building relationships with government customers is extremely difficult if you’re starting from scratch, but BAE Systems has enabled us to get our message to the relevant people. It’s like a stamp of approval that provides assurance to customers that we have a certain level of credibility.”
A selective approach
As mentioned at the beginning of this article, Fuzzy Labs is looking for interesting projects that enable it to have a positive impact. The focus isn’t on growth at all costs, so it wants to collaborate with other companies that align to that value.
To this end, the Fuzzy Labs team admit to being quite selective over who they work with, looking to establish strategic collaborations rather than prioritising quantity. For example, alongside the cultural element, organisations must also show customer/market alignment along with a complementary technical skillset and corresponding level of technical capability.
“We’re quite particular in what we do and how we do it, so we want the organisations we work with to reflect that. There’s no point us working with someone who can’t talk the same language as us.”Tom Stockton co-founder and CEO, Fuzzy Labs
The good news is that BAE Systems ticks all these boxes. What’s more, Tom explains how it has demonstrated an understanding of Fuzzy Labs as a business: “The collaboration with BAE Systems has been great for many reasons, one being that the opportunities the team has brought to us have been things we’re genuinely interested in. They’ve shown they get us. They’ve not just helped us commercially or on the clearance journey; the technology piece has also been really positive.”
But of course, these relationships can never be one-way. There must be some mutual benefit – whether that’s a financial incentive or access to something new. In this regard, Fuzzy Labs certainly holds up its end of the bargain by providing new specialist skills and expertise that can be deployed to help tackle the challenges government customers are looking to solve.
Onwards and upwards
So, this strategic relationship between a British prime and a fast-growing SME with niche expertise has already delivered value for both parties. But what does the future hold for Fuzzy Labs and BAE Systems?
Tom points to a desire to develop a joint future vision for the national security domain where Fuzzy Labs has quickly built a strong reputation and recognises an opportunity to bring a fresh perspective. “With fresh eyes, we look at the government sector and see that it’s been doing things the same way for a long time. We think we can help the sector try something new and bring added value. We want this collaboration to stand out and do something a bit differently.”
The opportunities are certainly there. What’s clear is the joint value collaborations such as these can deliver – a small organisation’s ability to iterate quickly and a large organisation’s depth of expertise combining to deliver outcomes that benefit customers and, ultimately, society at large.
Subscribe to The Digital Thread to hear more about the perspectives and experiences of our clients and stay up to stay with the latest insights from our experts