ElasticON London 2026: Interview with Elastic's Tim Brophy

Rapid AI innovation and adoption, shifting security and regulatory demands and the move from siloed, fragmented data to a unified, integrated or connected data ecosystem are some of the key challenges facing global businesses.

Elastic, the search AI company that enables organisations to securely find, analyse and visualise unstructured data, is at the forefront of solving these challenges. 

Its solutions help organisations cut through the data noise to find actionable, measurable answers and support innovation in three areas: enterprise search, observability and security. 

At Elastic{ON} London, the company’s annual gathering of developers, architects and business leaders, these themes were at the fore. 

The event, which took place on 26 February 2026 focussed on the transition from passive AI to agentic AI and how businesses can leverage Elastic’s solutions to enhance success.

Ahead of the event, Elastic’s Principal Solutions Architect, spoke with FinTech magazine. He discussed the challenges companies in financial services face, why AI is critical for innovation and how Elastic’s solutions support customers. 

Can you give me a brief introduction of what Elastic does?

Elastic is fundamentally a data platform built originally as a search engine, but which has been enhanced over time to support many different types of data –  time series data, telemetry emitted by machines and content generated by applications and people.

We take that underlying data platform capability and present it in three separate solutions across enterprise search, end-to-end observability and security.

Why is Elastic important FSI companies and what key challenges does it address?

FSI organisations typically have very distinct silos of data. Many of the traditional banks operate platforms that were essentially built in a different era, and not designed for today’s challenges around flexibility of data, speed of query and optimisation around asking questions of the data.

Banks are broken into personal banking, retail banking, mortgages, savings, loans and insurance. Not only are there data silos, but there are also application system silos, and it becomes really hard at a macro level within a bank to understand how applications are performing for end users – internally they’re viewed as separate things, but the consumer sees ‘one bank’.

Elastic is a perfect abstraction and unification layer that brings data from all of these different sources, normalises it across a standard format, and makes it easy to ask questions of that data, visualise it and run machine learning and anomaly detection on top of it. 

We have the capacity, the scale and the performant platform to do all of this in real time, which is really revolutionary.

Why does data hold such transformative value for AI, and why is the shift from siloed to unified data so important?

The demands of customers, regulators, platform owners and cost efficiencies are driving change in banks. 

The data volume is also exploding — we're at a point now where it's effectively too much for systems designed for resilience and atomicity of transactions.

The challenge is deriving value from AI, but that is very dependent on data. Banks need to unify that data in order through a solution like ours to truly derive value from your AI implementation. 

Gartner is predicting that by next year, 40% of all AI projects started in the last 3-4 years will be cancelled because they're not delivering on the value – they can't be trusted enough, and from an auditing and regulatory point of view, they just can't be explained well enough.

How does Elastic enable  hyper-personalised experiences?

Think of a case where you're searching for a policy. We can send that query to Elastic and turn it into what we call vector embeddings. 

We can search not only the words, but also the meaning of those words, can carve out the specific role a user has and the region that they work in, and find the most relevant document. 

If it's a five-page document but paragraph 10 contains the answer, we pull paragraph 10 and feed it back to a model to summarise – personalised to the user, translated if they're from a different region, and applicable to where they are from a regulatory point of view.

We also have machine learning models that can rerank search results based on what we understand about that user's demographic and push the most relevant results to them.

It’s a space that’s moving fast. What key trends are on your radar?

The trend that Elastic sees – and that I can see becoming more and more important – is context engineering. 

That means, how much refinement and relevance can we build into the mechanisms of providing the right level of information to a model so it can be useful? 

And then how much governance can we apply – not in a restrictive way, but in a way that helps us understand how a non-deterministic model came up with a particular answer. 

We can provide a mechanism to track end to end what happens before and after that model response, and test how truthful that answer was.

Context engineering is a significant trend going forward. The frameworks that AI is being implemented in are getting more and more sophisticated. And evaluation is going to become an absolutely critical aspect of every production workload in future.