From changing job roles and responsibilities to disruptive startups redefining entire industries. But is it true? Perhaps, said Brian Herron, director at user experience (UX) specialist Each&Other. AI was having an impact, he said, but it was also creating significant opportunities for incumbent players to take advantage of their large datasets and experience with customers. “With AI, we’re really starting to see some really considered movement,” he said. “As new models became available, the space got crowded. Now, we’re at a turning point in the hype cycle: a lot of the really interesting stuff you’re seeing now is companies that have been around for 10 or 15 years and have good structured data. They've been experimenting and now they’re starting to introduce AI features that aren’t just little tricks,” he said. To this end, Each&Other has partnered with PhasedAI, which specialises in developing and implementing robust AI solutions for businesses. “Through it [the partnership] we’re matching our deep product know-how with their deep technical skills,” said Herron. He explained that if companies have an installed user base for their product and good data, they now have challengers coming in, bringing excitement, investment, significant PR, and promising many big things. Nevertheless, incumbents possess an unparalleled understanding of their existing customer base and the problems their products solve. “You, as an incumbent, have to start thinking about refreshing and going through app modernisation, but also rethinking platforms around AI. The really huge advantage for the incumbents, though, is that they know their customers, and they know how people are using the apps and services,” he said. This has the practical upshot that they can integrate AI in ways that genuinely enhance user experience and address real customer needs, rather than just adding superficial features. “If they [incumbents] can move quickly, if they can update in a coherent way, and introduce useful AI features, not just magic tricks, but get to the heart of the job people use their products for. They can solidify their market leadership,” Herron said. Each&Other is already working with clients in areas as diverse as education, finance and market research. Getting things to the proof of concept stage is generally not the hardest thing in the world, but to get to where you are fully operational, there is a risk “These are digital companies and they’ve been running their products successfully for years, so they do have capable internal teams. The question is: are they capable of doing these new things and, even if they are, is there an opportunity cost to moving away from what they are doing,” Herron said; “PhasedAI is working natively in the [AI] space, addressing complex issues like the tech stack and architecture, addressing the risk around hallucination. You end up, then, with the company bringing its experience in the market; we’re bringing in expertise in terms of product strategy and then PhasedAI is bringing in its really deep knowledge of AI,” he said. The point, Herron said, was not to overwhelm companies with a complete overhaul, but instead to enable them to strategically integrate AI to achieve tangible business outcomes. This often involves a phased approach, starting with high-impact, low-risk AI features and scaling up as the organisation gains confidence and expertise. Put simply, the goal is evolution through integration, not revolution through disruption. “We can bring companies to whatever stage they need to get to, and then their team can take over,” he said. Ultimately, the success of AI adoption for incumbents hinges on a pragmatic approach to data and implementation. Indeed, data is the foundation that can make or break any AI transformation effort. However, beyond just having data, the crucial step is ensuring it’s clean, well-structured, and accessible for AI models to learn from effectively. This often requires significant upfront work, but it’s a non-negotiable step for those wishing to harness AI’s full potential. “If there’s a data problem it’s enormous. Getting things to the proof of concept stage is generally not the hardest thing in the world, but to get to where you are fully operational, there is a risk. You could waste resources, and it could blow up in your face,” Herron said.