Don't Use AI for Everything: Elastic on Search as the Context Layer | Money20/20 Europe 2026

Quick Summary

At Money20/20 Europe 2026, Arno van de Velde, Principal Solutions Architect at Elastic, sets out the balance financial institutions are still getting wrong with AI. The opportunity is obvious and the temptation is to apply AI to everything, including the parts of a solution it is least suited to. His answer is to use search as a context engineering layer, so the model is given the right information rather than asked to supply it.

What does Elastic do in financial services?

Elastic is a search AI platform company, and van de Velde describes a footprint across search, observability and security, alongside other technologies, inside banking and finance. That combination matters in a sector where the same underlying data has to serve customer-facing search, system monitoring and threat detection at once.

Where do banks go wrong with AI?

The biggest opportunity right now is incorporating AI, and van de Velde is clear that the accompanying challenge is using it indiscriminately: applying AI to everything, and to the wrong parts of a solution. Used as a context engineering layer, search is how he sees institutions balancing those competing pressures, retrieving the right material so AI is applied where it genuinely adds value.

FF NEWS TAKE:

This is the most useful correction in enterprise AI at the moment, and it is notable that it comes from a search vendor rather than a model vendor. Most failed banking AI projects are retrieval problems wearing a model-shaped costume: the answer is wrong because the context was wrong. Framing search as context engineering rather than as a feature is the right way round, and it puts the unglamorous data work back where the value actually sits.