Xebia: The Crucial Importance of Data for Agentic AI

In the field of artificial intelligence, the importance of a robust data foundation cannot be overstated. Niels Zeilemaker, Global CTO at Xebia, highlights that to effectively integrate AI agents into an organization, it is crucial to make data accessible and usable by these agents. According to him, even the best AI agent will not function properly if it does not have access to the right data. It could misinterpret information or make incorrect connections between different fields of data, which is not necessarily due to a failure of the agent, but rather to an inadequate data foundation.

Zeilemaker emphasizes the importance of data cataloging. Although this concept is not new, it takes on a different dimension with AI agents. Unlike humans, who can compensate for a lack of documentation through informal exchanges, AI agents must rely entirely on the data catalog. If a description is incorrect or incomplete, the agents will not be able to operate optimally. This underscores the need for precise and comprehensive documentation to avoid costly errors. Zeilemaker stresses that agents must rely on the data catalog and what is written there – and if the description is incorrect, the agents will not perform.

Xebia strives to transform companies' AI strategies into production-ready solutions, enabling rapid and efficient transformation. The company prides itself on adhering to core values such as respect for people and uncompromising quality. However, Zeilemaker points out that knowledge sharing may be the most crucial aspect. At events like TechEx Global North America, Xebia shares its expertise, allowing it to stay at the forefront of technological innovations and quickly adapt to market changes.

At the AI & Big Data Expo, Zeilemaker presented Xebia's Agent Data Foundation (ADF). This platform is designed to host AI agents and unify fragmented data landscapes within companies. By combining specially designed AI agents with expert engineering, Xebia significantly reduces the migration time to modern platforms, compressing a 12 to 24-month timeline into a fixed-price engagement with defined milestones. After performing migrations in the traditional manner and accelerating some with LLM coding, Xebia is now integrating this into the data platform, leveraging the additional context it can provide to further accelerate migrations.

A central element of this strategy is what Xebia calls the Agent Data Foundation (ADF). This approach allows for the extension of the data platform to host agents, which are then used in customer-oriented use cases and internal processes. While migrating from legacy platforms to modern platforms has always been a priority, Xebia is seeing a growing demand for faster and more reliable migrations. Zeilemaker explains that this demand is met through close collaboration between the consultant and the client to co-develop the solution.

Xebia also offers Xebia ACE: AI-Native Software Engineering, a framework that integrates AI throughout an organization's software development lifecycle (SDLC). This approach enables project delivery to be accelerated by up to 40%, while reducing legacy system transformation costs by up to 70%. Zeilemaker emphasizes that Xebia ACE is particularly beneficial for large enterprises that wish to maintain strict governance while adopting modern development practices. Zeilemaker notes that Xebia ACE is especially useful for large companies that "perhaps still want to adhere to a particular governance or way of working while doing SDLC."

A concrete example of the impact of Xebia ACE is vibe coding. While anyone can create applications, few dare to put them into production without a reliable framework. Xebia ACE provides this security while maintaining high quality. Zeilemaker explains that adopting ACE allows organizations to benefit from the advantages of LLM acceleration without sacrificing the quality of the final results. Zeilemaker uses vibe coding as an example.

Finally, Zeilemaker addresses the importance of control in AI-driven software development. With the increase in AI-generated code, the SDLC could become a security weakness. For companies, this control is essential. With so much code generated, AI-driven SDLC could become a security vulnerability through weaknesses. He notes with interest Anthropic's initiative to introduce a pull request reviewer, adding a layer of review by an LLM, which could become a standard in the industry. This approach would allow for third-party review by a very senior team member in the form of an LLM, thus ensuring increased security during production releases.

In conclusion, no matter where organizations are in their journey, from assessing their data readiness to building, Xebia is capable of helping to establish the right foundations and create the resulting transformations. The experience accumulated by Xebia in this field has led to the development of Xebia Axis: Agent Data Foundation, a solution that helps companies make their data AI-ready faster than any other alternative.