From AI hype to AI help

Australian organisations are pouring millions into artificial intelligence, yet most are struggling to turn experimentation into real-world impact.

“Boardroom AI projects aren't failing because the LLMs are flawed, they’re failing because they are data-blind. Without the right context, agents guess. And a guess in an enterprise setting is a hallucination that wastes millions.”

That’s the blunt assessment from Jeremy Pell, ANZ country manager at Elastic, and it’s a message set to anchor the company’s upcoming ElasticON Sydney conference on March 5, 2026. 

According to KPMG, AI has surged from fourth place to become the number one challenge facing Australian executives in just 12 months. But despite soaring boardroom interest, the reality on the ground tells a different story.

“When I speak to Australian executives, I'm hearing a common frustration: they’ve invested millions in AI, yet according to MIT, 95% of these projects are failing to move past the pilot phase,” Pell says.

Elastic will bring that disconnect into sharp focus at its upcoming Sydney conference, Elastic{ON}, part of a global developer-focused event series expected to attract 400–500 engineers, architects, analysts and platform leaders locally.

The event is aimed squarely at the practitioners building, securing and operating modern systems: the people “at the coalface” of AI adoption across search, observability and security.

Why enterprise AI keeps stalling

Pell says most enterprise AI initiatives aren’t failing because large language models are flawed.

“Elastic provides the eyes by grounding agents in actual enterprise data, ensuring every response is a fact rather than a creative guess.”

That problem is compounded by the way corporate data is typically stored.

“Most of a company's gold is trapped in messy unstructured data such as PDFs, emails, and technical logs, that traditional systems can't easily read,” Pell says. “Elastic specialises in searching this unstructured world, allowing agents to instantly find that one specific email thread or PDF clause needed to solve a customer's problem.”

The shift, he argues, is moving away from prompt engineering toward what many now call context engineering, grounding AI agents in live enterprise data so responses are factual, traceable and actionable.

From experimentation to execution

Certainly, the challenge facing many Australian organisations is no longer whether to adopt AI; instead, it’s how to make it reliable, secure, and economically defensible.

With agentic AI gaining momentum, companies are racing to deploy autonomous workflows across customer service, IT operations, security and development pipelines. But Pell warns that rushing ahead without the right data foundations is leading to stalled projects and missed ROI.

“To build a reliable agent, you need the full picture of your data assets. Elastic is unique because it allows you to search across search, security, and observability data in a single unified platform. You don't have to move or duplicate data, you simply give your agent the ability to see across the entire estate to find the answers that matter.”

Elastic recently introduced new agent builder capabilities and workflow tooling designed to help organisations operationalise AI across distributed environments, an approach Pell says is essential as enterprises attempt to scale beyond proof-of-concept deployments.

Practical lessons for builders

Elastic{ON} Sydney will showcase real-world use cases spanning application development, platform reliability, threat detection and operational visibility, with sessions designed to help technical teams translate AI ambition into working systems.

Pell says the focus this year is firmly on outcomes.

“Your leaders are looking for answers, and they are looking to you to build them. We’ve entered a new era where we have to move from AI hype to AI help. The days of doing AI for AI’s sake are over. Leaders are demanding a real business case and tangible ROI.

“At Elastic{ON}, we’ll be helping teams on the ground turn AI ambition into practical, production-ready systems.”

Register now for Elastic{ON} Sydney on the event website.