Big Data LDN 2025: The Fast-Evolving World of AI

Big Data LDN is definitely one of the most exciting events in the calendar for anyone who is interested in data, and this year was no different.

From data governance to sustainability and Data For Good, the topics at this year’s event were really interesting, but nothing could compare to the wave of excitement around Gen AI.

Attending BDL 2025 was a reminder of just how fast the world of AI is evolving. Over the past few years, we’ve seen AI accelerate not just task automation, but the accessibility of once highly technical and specialised processes.

What was once reserved for data scientists, engineers, and analysts are now increasingly within reach for a much broader range of users.

One of the standout sessions for me was a talk by Tammie Coles (CData Software) and Sami Hero (Ellie AI) titled: “Bringing Data Modelling to the Masses with AI and Embedded Connectivity”. Instead of starting from scratch or relying solely on analysts or engineers to interpret requirements, Ellie connects directly to existing systems, pulls in metadata, and uses AI to suggest entities, relationships, and definitions.

This provides a visual, accessible environment where business stakeholders can recognise their own terms (“tenant,” “property,” “maintenance request”), and engineers can see how those concepts map to real data structures.

Furthermore, by embedding connectivity through partners like CData, Ellie reduces the barriers to entry: users don’t have to wrangle complex connectors or ETL jobs just to understand what data exists.

A particularly thought-provoking moment came during the Q&A, when an audience member asked Sami whether AI could build predictive data models that show the impact of changing a company’s data infrastructure.

Sami’s answer was illuminating. He explained that while AI can help generate a "future" model, compare it to the current one, and surface insights without needing to build the full functionality, the creation of that future model still requires human input and strategic thinking. In other words, the automation can go far, but the imagination behind the design still needs to come from us.

I think what would be exciting is if we could prompt AI into building a model with specific impact criteria whilst utilising tools such as Ellie. If we could combine Ellie with other tools like Sifflet and tell AI: “Design a new tenant services model, but don’t disrupt how we track statutory safety checks, make sure rent arrears reporting remains untouched, and ensure maintenance workflows still run within their current service-level targets.”, that would be really powerful.

Ellie could generate the draft model, while Sifflet’s lineage and impact analysis would simulate downstream effects and highlight risks. Together, this could create a feedback loop: design → assess → refine → approve, ensuring models evolve quickly but safely within organisational guardrails.

Big Data LDN once again was a thought-provoking event that has given me a lot to think about, especially when it comes to the ever-changing world of AI.

Want to chat about how Dot Collective can help with your latest AI project? Get in touch.