What we heard at the Gartner Data & Analytics Summit 2025

1. GenAI is crashing back to earth

Remember when every second conversation was about building a chatbot or automating workflows with GenAI? That excitement hasn’t gone away, but it’s getting a lot more grounded.

Businesses are starting to ask harder questions:

• Are we seeing real ROI?
• Is the output good enough?
• Is the cost worth it?

The shiny demos are fading. What’s left are the use cases that actually solve problems

2. Everyone wants “AI-ready” data—but few have it

A recurring theme across the summit: your data needs to be ready before your AI can be useful.

That means:

• Clean, structured, documented datasets
• The right metadata and governance
• Strong pipelines and architecture

According to Gartner, the number one investment priority for data leaders over the next 2–3 years is building AI-ready data foundations. It’s not glamorous work, but it’s the difference between a GenAI project that works, and one that ends up on the shelf.

3. Small models are winning

The narrative around LLMs is shifting. Instead of training massive general-purpose models, many teams are now building smaller, task-specific ones, which are faster to deploy, easier to control, and way more cost-effective.

Gartner predicts these smaller models will be used three times more than big ones by 2027. For most organisations, that’s a clear signal: don’t overbuild. Focus on solving one problem well.

4. Metadata is having its moment

Metadata isn’t the most exciting thing to talk about—but at this year’s summit, it was everywhere. Why? Because it’s the foundation for everything AI touches. Prioritising metadata and semantic context can improve GenAI model accuracy by up to 80% and cut costs by 60%. 

5. AI should feel boring

One of the best quotes from the event: “AI should make your life easier, not more complex.”

The point? Simplicity wins. Organisations are realising that the best AI outcomes come from:

• Fewer steps
• Less friction
• More automation behind the scenes

Think: invisible agents that summarise meetings, recommend next steps, or reduce repetitive tasks. Not headline-grabbing. Just helpful.

In short: less hype, more substance

This year’s summit felt different. There’s still ambition – but there’s also a shift toward focus, governance, and delivery.

• If you’re building something with AI or data, the message is clear:
  Start with strong data foundations
  Solve real problems, not hypothetical ones
  Measure value like you mean it

Need help figuring out where to focus? That’s what we do 😉