The second day of the AI & Big Data Expo 2026, held alongside Digital Transformation Week in London, revealed a clear evolution in enterprise AI adoption. The initial excitement surrounding generative AI models is giving way to practical challenges faced by organizations integrating these technologies into established infrastructures.
Transition from Generative Models to Infrastructure Focus
While early discussions centered on the promise of large language models, day two shifted the spotlight to the foundational elements necessary for successful AI deployment. These include data lineage, observability, compliance, and the critical need for robust data management practices.
Data Maturity as a Cornerstone for AI Success
Experts stressed that AI reliability is fundamentally tied to data quality. DP Indetkar of Northern Trust cautioned against the risks of deploying AI with poor data inputs, likening such scenarios to a “B-movie robot” failure. He emphasized that organizations must achieve analytics maturity before embracing AI-driven decision-making, as automated systems can amplify errors if fed unreliable data.
Supporting this view, Eric Bobek from Just Eat highlighted the importance of a strong data foundation, warning that investments in AI layers are futile if underlying data remains fragmented. Similarly, Mohsen Ghasempour of Kingfisher pointed out that reducing latency between data collection and actionable insights is essential, especially for retail and logistics sectors aiming to realize tangible returns.
Scaling AI in Highly Regulated Industries
The finance, healthcare, and legal sectors present unique challenges due to near-zero tolerance for errors. Pascal Hetzscholdt from Wiley underscored that responsible AI in these fields demands accuracy, clear attribution, and system integrity. He stressed that auditability is non-negotiable as “black box” AI implementations risk reputational damage and regulatory penalties.
Konstantina Kapetanidi of Visa discussed the complexities of building scalable, multilingual generative AI applications that actively use tools like databases. This approach introduces new security vulnerabilities requiring rigorous testing to prevent potential breaches.
From Lloyds Banking Group, Parinita Kothari challenged the notion of “deploy-and-forget” AI systems, advocating for continuous monitoring and maintenance akin to traditional software infrastructure to ensure ongoing performance and compliance.
Transforming Developer Workflows with AI Copilots
AI is reshaping software development processes. A panel featuring representatives from Valae, Charles River Labs, and Knight Frank examined how AI copilots accelerate code generation but simultaneously increase the emphasis on code review and architectural oversight.
Further discussions involving Microsoft, Lloyds, and Mastercard highlighted the skills gap in the current workforce, urging organizations to invest in training programs that enable developers to validate AI-generated code effectively.
Dr. Gurpinder Dhillon from Senzing and Alexis Ego from Retool showcased how low-code and no-code platforms integrated with AI can expedite the delivery of internal applications. This strategy aims to reduce the backlog of tooling requests without compromising quality, offering cost efficiencies for enterprise software development.
Adapting Workforce Models and Enhancing Utility
The broader workforce is beginning to collaborate with “digital colleagues,” AI agents that actively participate in workflows. Austin Braham from EverWorker emphasized the need for businesses to redefine human-machine interaction protocols to accommodate these changes.
Paul Airey from Anthony Nolan shared a compelling example of AI’s life-saving potential, detailing how automation improves donor matching and transplant timing in stem cell therapies.
A consistent theme was that successful AI applications often target specific, high-impact problems rather than pursuing overly broad solutions.
Managing the Transition to Production-Ready AI
As enterprise focus shifts from novelty to integration, priorities now include ensuring uptime, security, and regulatory compliance. Organizations must strengthen data infrastructure, establish legal frameworks, and train personnel to supervise AI-driven systems.
Executives are advised to channel resources into data engineering and governance to avoid pilot projects stalling due to inadequate foundations. The difference between successful AI deployments and abandoned initiatives often resides in these fundamental preparations.
Related Articles
Delve Faces Allegations of Misleading Customers with False Compliance Claims
Apple Assigns AI Content Disclosure to Music Labels and Distributors with New Transparency Tags
Sierra, Bret Taylor’s AI Startup, Surpasses $100 Million ARR in Under Two Years
Mistral AI Unveils Mistral 3: A New Open-Source Multilingual and Multimodal AI Model Suite
