At this year’s Oxygenate event, WSO2 explored how the theme of “AI Meets Open Source” is driving a new era of enterprise transformation. From agentic AI that boosts developer productivity to localised LLMs ensuring trust and governance, Isabelle Mauny, Field CTO at WSO2, explores how WSO2 is helping organisations move beyond experimentation to real-world impact, unlocking the true potential of open, responsible AI.
What excites you most about this year’s Oxygenate theme – AI Meets Open-Source?
AI is currently a major focus for our customers, and we are concentrating on helping enterprises move beyond experimentation to build real, value-driven solutions. Integrating AI across our offerings has fundamentally transformed our open-source products. I am particularly excited about the insights, demonstrations and product showcases we’ll be presenting this year, which will illustrate how AI is reshaping what’s possible in the open-source landscape.
Where does AI genuinely add value to enterprises?
The immediate value lies in productivity gains. Developers are already under significant time pressure, so even a 10–20% improvement in task efficiency represents meaningful progress; in other words, time they can reinvest in higher-value work.
The second dimension is the rise of agentic AI. Businesses now have the opportunity to offer entirely new ways for users to interact with their systems through AI agents and natural language interfaces. To use an extreme but plausible example, we may eventually move beyond the need for traditional mobile apps, instead engaging through a single intelligent interface that can answer questions and take action.
This shift will fundamentally change how we work and how enterprises think about customer experience. Today, most of the value is still on the productivity side, as we continue strengthening the foundational elements, reliable architectures, strong guardrails and trustworthy frameworks, for agentic AI to operate safely and effectively.
How do you see the relationship between governance, trust and security evolving in the open-source landscape?
The challenge does not stem from developers, but rather from non-technical users (e.g. those in HR, legal or other business functions) who are now using generative AI tools like ChatGPT to summarise documents or create presentations. Ironically, these are people who would never share such sensitive information with another person yet feel comfortable uploading it to an AI system. We must therefore operate under the assumption that everything written or shared could be analysed and stored somewhere.
The key questions become: How do we safeguard our data? How do we prevent misuse? AI models are designed to help make them susceptible to being manipulated into revealing sensitive information. This is why we expect to see a growing shift toward localised LLM deployments: domain-specific models trained on private data that do not have access to everything.
However, this decentralisation introduces a new challenge: the proliferation of models that all need proper governance. We must ensure visibility and control: who is using which model, what kinds of requests are being made and what the consumption levels are. Enterprises need to avoid paying for thousands of employees using AI for personal queries. These issues highlight the urgent need to establish governance frameworks now, rather than repeating past mistakes.
As the Field CTO, you’re at the intersection of customers and the product. What is a recurring challenge that you’re hearing from customers right now?
The most common challenge is staying competitive. There is less need for hard selling now, as most enterprises already want to adopt AI solutions. The shift has been from us approaching customers with, “How can we help you?” to them approaching us asking, “How can you help me?”
At a recent Oxygenate event in Italy, several participants expressed exactly this concern. They see competitors publicly announcing new AI-driven initiatives and recognise the need to respond, but they often don’t know where to start. That uncertainty is a recurring theme across industries.
Looking ahead, what are you preparing for and investing in as a priority at WSO2?
We are heavily investing in agentic AI assistants, i.e. intelligent companions that work alongside developers to ensure their source code, APIs and configurations are of the highest quality and security, while also automating certain tasks. This technology is especially valuable for new customers, helping them get up to speed faster and use our products more effectively.
Another major milestone is WSO2’s recent acquisition of Moesif, an analytics and monetisation platform. This addition enables us to aggregate data from multiple partner applications into a single, unified sales interface, offering powerful insights and new opportunities for growth.
Finally, customer support remains a key focus. Approximately 20% of our support queries are now answered automatically by AI agents with a very high satisfaction rate. While these are not front-line product questions, they represent an important step forward in efficiency and responsiveness, and we are continuing to invest significantly in this area.
