Skip to main content

By Oliver Saywell, Head of Databricks at Nimble Approach

Last week, we were proud to sponsor the Databricks Data + AI World Tour in London at ExceL. It was a day packed full of conversations, demos, and practical discussions with data and AI leaders, practitioners, and decision-makers from across the UK and beyond. Having a stand meant we spent the day in direct dialogue with clients and prospects, and the themes that came through loud and clear centred on two tightly linked priorities: 

  • How organisations can extract real value from Genie and broader AI capabilities.
  • Why strong data foundations remain non-negotiable.

The energy on the floor

The London stop reflected the wider World Tour focus: moving beyond AI experiments to production systems that deliver measurable outcomes by grounding models and agents in enterprise data and business context. Sessions and conversations repeatedly returned to the same practical questions: 

  • How do we make AI trustworthy and useful for the business? 
  • How do we scale it without creating new silos or governance headaches? 
  • How do we ensure the underlying data platform can actually support agentic workloads?

For the public sector leaders we spoke with, these questions carried a unique weight. Historically, government has treated safety and speed as a trade-off: moving fast meant risking compliance, while tight governance often meant glacial delivery. But a standout theme from the Public Sector Keynote was that modern data foundations are finally helping to eliminate that compromise. Whether stress-testing special educational needs policy at the Department for Education or putting live intelligence into the hands of frontline responders, government is demonstrating that it is possible to optimise for safety and speed simultaneously.

Our stand was a hub for these discussions. Clients and attendees wanted concrete examples rather than high-level vision, and many were already experimenting with or evaluating Genie. Across these conversations, one thing was clear: the appetite to move from AI potential to practical, governed and measurable impact is growing.

Getting more out of Genie and AI

Genie featured heavily in our conversations. The latest generation positions Genie as a more autonomous, always-on AI coworker that can handle longer-running work, operate within organisational guardrails, and draw on richer business context. Attendees were particularly interested in:

  • Moving from pure conversational analytics to agents that can take ownership of multi-step tasks and automating business processes while still escalating decisions to humans when needed.

  • Using Genie Ontology and related context layers so answers and actions reflect the organisation’s actual definitions, metrics, relationships, and knowledge, not generic LLM knowledge.

  • Combining structured lakehouse data with unstructured sources (documents, knowledge bases, enterprise apps) in a governed way.

  • Practical adoption patterns: starting with well-scoped domains or teams, measuring trust and accuracy, and gradually expanding as confidence grows.

From a client perspective, the opportunity is clear. Genie and related AI capabilities can dramatically lower the barrier for business users to access insights and automate routine work. But the conversations also highlighted the gap many organisations still face: exciting demos do not automatically translate into reliable, scalable production use. Success depends on deliberate design of the experience, clear ownership, continuous evaluation of answer quality, and tight integration with existing processes and permissions.

In public services, this human-centred approach is non-negotiable. Whether a policy lead is stress-testing legislation with Genie or a local authority is connecting housing and benefits data to help prevent homelessness before crisis hits, AI only delivers value when it puts trusted intelligence into the hands of the right person, at the moment they need it, without removing human accountability. The challenge, then, is turning that potential into practical, measurable outcomes.

Many of the most productive discussions focused on pragmatic next steps, identifying high-value use cases where Genie can deliver quick wins, defining what “good” looks like for accuracy and auditability, and ensuring the right people (business + data + platform teams) are involved from the start.

Data foundations still decide how far AI can go

Almost every conversation about Genie or agentic AI eventually circled back to the data foundation. This was not theoretical. Clients repeatedly described the same constraints: inconsistent definitions, fragmented ownership, gaps in governance and lineage, and data that is not yet reliable or discoverable enough for agents to trust.

Key points that resonated:

  • Unity Catalog and the lakehouse architecture remain the practical foundation for governed, discoverable, high-quality data that AI systems can safely use.

  • Semantic consistency and business context (metrics, hierarchies, domain knowledge) need to be treated as first-class assets and not afterthoughts.

  • Data quality, freshness, and access controls directly determine whether Genie and other AI tools produce answers people will actually rely on.

  • Organisations that invest early in solid foundations find it far easier to scale from pilot Genie Spaces or agents into broader production use.

  • Openness and sovereignty are particularly important for public sector bodies, with open data formats like Delta Lake and Iceberg enabling agencies to break down departmental silos and share data safely across boundaries, without locking the state into proprietary vendor ecosystems.

In short, AI ambition is only as strong as the data platform underneath it. The most forward-looking clients we spoke with are treating data foundation work and AI adoption as a single programme rather than sequential phases.

What this means for organisations right now

The London event reinforced a simple message: the technology is advancing quickly (Genie being a clear example), but sustainable value still comes from disciplined execution. Organisations that will remain ahead are those that:

  1. Pair AI exploration with deliberate investment in data quality, governance, and business semantics.

  2. Start with focused, high-impact Genie or agent use cases and measure outcomes rigorously.

  3. Bring business users into the design process early so the resulting experiences solve real problems.

  4. Treat the platform (lakehouse + Unity Catalog + AI capabilities) as a strategic asset rather than a collection of tools.

Looking ahead

Sponsoring the event and staffing the stand gave us a valuable window into where clients are today and where they want to go. The appetite for practical guidance on Genie, agentic workflows, and the underlying data foundations is high. If you were at the event and we didn’t get the chance to connect properly, or if these themes resonate with challenges you’re facing, we’d be glad to continue the conversation.

The Data + AI World Tour London was a strong reminder that the most interesting work sits at the intersection of powerful new AI capabilities and the unglamorous but essential work of making data trustworthy and accessible. That’s exactly where we focus and we’re excited to help more organisations turn the potential we saw in London into results.

Continuing the conversation at Databricks North Connect

The conversations we had in London are part of a much wider discussion about how organisations can make the most of their data and AI investments. There are plenty more insights and experiences to share, and we’ll be continuing that conversation at Databricks North Connect in Leeds this December.

Bringing together the Northern data community, it’s a great opportunity to connect, exchange ideas, and explore the latest developments across the Databricks ecosystem.

Join us at Databricks North Connect on 10th December. Find out more and register here.

Get In Touch