AiThority Interview With Claire Southey, Chief AI Officer at Rokt

Claire Southey, Chief AI Officer at Rokt chats about the future of AI and impact on software development processes in this AiThority interview:

_____________________________

Hi Claire, tell us about your journey in the tech space and more about being Rokt’s chief AI officer?

My journey in tech has always sat at the intersection of data, decision-making, and real-world impact.  I started out deeply technical, working with internet-scale systems and using data science to advance knowledge and understanding of how the world works.  Over time, I became increasingly focused on how data, especially predictive analytics, can actually advance and enrich the human experience, rather than just explain it.

Today, my role spans strategy, architecture, research and science, governance, and also culture – helping teams understand where AI adds value, where it doesn’t, and how to build it and use it thoughtfully. In my role, AI is not a side project or buzzword.  It is a core capability I embed into how our business builds products and makes decisions.  

What does a Chief AI Officer do in your terms?

The Chief AI Officer is still an emerging role, and I see it as one of stewardship as much as innovation. At its core, my responsibility is to turn data and AI into real business leverage. That starts with setting a clear AI and data strategy that’s tightly aligned to business outcomes, and making sure we have the right datasets, infrastructure, talent, and foundations to support it. It also means guiding our scientific research efforts, both internally and through partnerships with academia, while ensuring our models and data pipelines are reliable, ethical, and privacy-safe. We design and train our own proprietary models where it gives us a real advantage, and we leverage proven third-party models where it doesn’t make sense to reinvent the wheel; part of my job is helping teams make that call thoughtfully. Ultimately, my goal is to build platforms and tools that make data and AI genuinely accessible and useful across the entire organization.

Also Read: AiThority Interview with Zohaib Ahmed, co-founder and CEO at Resemble AI

How is AI playing a key role in modern eCommerce? Some highlights around how leading brands are using AI to drive impact?

Consumers are more empowered than ever, with instant access to vast product catalogs, reviews, and price comparisons across countless channels.  However, the ‘paradox of choice’ is that offering consumers too many options increases decision fatigue and reduces the likelihood of purchase. High quality product recommendations solve this by narrowing the field to a small, relevant set of options that match the shopper’s intent, making decisions easier and more likely to convert. High quality recommendations and cross-sells are critical to competitive advantage.  Today, 35% of Amazon’s revenue comes from personalized recommendations, and 80% of Netflix’s viewing comes from content recommendations. In this environment, AI becomes a core source of competitive advantage: the brands that win are those that can best leverage high-quality historical interaction data, continuously capture new behavioral signals, and apply advanced machine learning techniques to deliver more relevant, personalized experiences than their competitors.  At Rokt, we have been at the forefront of this technology for more than a decade, continuously raising the bar on the online shopping experience and delivering stronger outcomes for our clients.

Tell us about some of Rokt’s latest AI enhancements and how they empower end users?

Our mission is to design and deploy AI and machine learning systems that make ecommerce more human for everyone.  That means recommending the best next steps for a consumer at any moment in their shopping journey.  We have the privilege of accompanying consumers through seven billion ecommerce checkout experiences every year, learning from those shoppers’ experiences, and finding ways to optimize and predict the best next step or product recommendation.  We know that understanding the full consumer context matters greatly, which is why we acquired mParticle for $300M USD last year.  We are already using advanced modelling techniques like entire-space, multi-task learning approaches, and our science teams are currently studying new developments around incrementality simulation and contrastive learning.  We’ve come a long way since we deployed our first AI predictive model back in 2015, and we are far from done.  But we don’t build tech for its own sake.  Every innovation, every model enhancement we engineer, ensures that shoppers get a better experience with products and services that delight them, and retailers get stronger engagement and higher purchase rates.

What thoughts do you have around the future of AI and how it will impact software development cycles and teams?

As AI gains deep, contextual understanding of entire codebases, software development will shift from a craft of construction to one of orchestration. Exponential advances in tooling will dramatically compress software development cycles, enabling smaller teams to build and evolve systems of unprecedented complexity.

We’re already responding to that change culturally. Last year, we compressed our planning cycles from quarterly to four-week horizons, because in an environment evolving this quickly, the bigger risk is over-planning and under-iterating.

Importantly, this doesn’t diminish the human role. It elevates it. As machines take on more of the mechanics of code generation and execution, teams will spend more time on direction, judgment, trust, and disambiguating requirements. The defining skill of future software teams won’t be how much code they write, but how clearly they express intent, move from ambiguity to clarity, and govern the impact of what they build.

It’s also changing how we think about technical debt. I’m more tolerant of it today than I was two years ago, because AI-driven code comprehension is advancing so rapidly that the cost of refactoring and cleanup is falling fast. That’s not a license for sloppy engineering; it’s a recognition that, right now, speed of learning and iteration often matters more than technical perfection.  I can afford a little more tech debt today because the cost of cleaning it up tomorrow will be much lower.

Some of the most exciting AI innovations that pique your interest from around the world before we wrap up?

While language models dominate attention today because of their reasoning capabilities at scale, some of the most interesting progress for me is happening in reinforcement learning, particularly around abstract reasoning. AI has historically excelled at automating well-defined tasks; the next shift will be toward systems that help explore and generate new knowledge. If 2025 was the year of agentic AI, 2026 may be when that potential starts to translate into generating original hypotheses, new algorithms, discovery of new knowledge, and solutions to long-standing unsolved problems in math, chemistry, medicine, physics, and economics.

Also Read: The Death of the Questionnaire: Automating RFP Responses with GenAI

[To share your insights with us, please write to psen@itechseries.com]