Article complete

Most AI Investments Will Fail, Unless…

In a study, about 50% of CEOs admitted that the recent surge in technology investments has resulted in disconnected systems, hindering their ability to build an effective data foundation

A stylized illustration of Benjamin Franklin covering his face with one hand, revealing "AI" in the eye space.

Key takeaways

  • Recognize that messy data is hindering AI investment returns for many organizations.
  • Adopt an integrated data architecture to enhance cross-functional collaboration and AI effectiveness.
  • Acknowledge that 68% of CEOs see enterprise-wide data structures as essential for AI success.
  • Prepare for rapid AI growth as 61% of CEOs are actively implementing AI solutions.
  • Understand that half of CEOs report disconnected systems due to recent tech investment surges.

Takeaways drafted with AI from the article text.

With companies racing to cash in on AI, CEOs are hitting a brick wall, and messy data is the main culprit. With data all over the map, AI’s true potential might get strangled at the source, thereby yielding not-so-desirable return on investment (ROI) on all AI investments.

An IBM study, which surveyed 2,000 CEOs worldwide, found that AI investments are poised for rapid acceleration. Executives expect the growth rate to more than double over the next two years. Currently, 61% of CEOs report they are actively adopting AI agents and are preparing to scale their implementation.

However, the study also highlights the growing importance of data in realising AI’s full potential. A significant 68% of CEOs consider an integrated, enterprise-wide data architecture essential for effective cross-functional collaboration, while 72% believe their organisation’s proprietary data is key to unlocking the true value of generative AI.

Despite this recognition, many organisations appear to be grappling with fragmented data environments. Half of the participating CEOs admitted that the recent surge in technology investments has resulted in disconnected, piecemeal systems, which hinders their ability to build a cohesive and effective data foundation.

What is the Problem?

In a recent interview with AIM, Alexander Rinke, co-founder and co-CEO of Celonis, highlighted the growing capabilities of AI models. “The problem is not the models (LLMs). The models are wicked smart already, and they're getting smarter every day.”

However, he pointed out that the real challenge lies in the unique process context of each organisation. Every company operates with distinct processes and data, which are often fragmented across multiple systems.

Organisations are increasingly using agentic frameworks to reduce workloads and boost productivity. These frameworks leverage intelligent agents to handle tasks, but each workload is triggered by a process. 

Therefore, if organisations can gain deeper insight and intelligence about the ideal version of a process, they can prevent the workload from rising in the first place. “We can provide the most important worker instruction detail for those agentic frameworks—for example, where to tackle the problem,” Rinke added.

Celonis has made significant strides in enabling this vision. They use AI to help ingest data from various systems into a unified model known as the process intelligence graph (PI graph). This graph represents connected processes across systems, offering a comprehensive view.

“The idea behind agency is that any agent framework can be used—whether it’s Amazon Bedrock, Microsoft Copilot Studio, or any other platform,” Rinke explained.

By exposing process context and intelligence through a single API, Celonis ensures that AI agents can access the precise information needed for execution. 

In addition to these foundational tools, Celonis offers other AI-driven products, such as Copilot and the machine learning workbench. The main advantage, as Thomas pointed out, is Celonis’ ability to fuel any type of agent with highly relevant intelligence about the processes of the company that’s unique to that company.

This perspective on process intelligence aligns with the views of Nitin Chandel, group VP and India country manager at UKG, who shared his thoughts with AIM. “As organisations increasingly embrace AI agents to drive efficiency and agility, process intelligence has become essential.”

According to Chandel, PI offers the contextual understanding that AI agents need to execute tasks effectively and understand how work flows across systems, teams, and processes. He compared agents without this context to interns without training—fast but clueless. “With it, they become informed, proactive, and far more valuable.”

Furthermore, Leslie Joseph, principal analyst at Forrester, told AIM that process intelligence is the basis for operational governance. It enables organisations to monitor, audit and course correct agent behaviour—all essential elements of trust.

Who Needs It?

Global capability centres (GCCs) have evolved from operational support units into strategic hubs driving transformation, analytics, innovation, and value across global enterprises. However, this shift brings a new set of expectations: to deliver faster, leaner, and smarter operations, often powered by AI, across global functions, all while keeping costs and complexity in check.

He further emphasised that while enterprise AI may be smart, it isn’t intelligent enough to understand the complete picture yet. “This is especially critical for India’s GCCs. They are running complex global processes across functions and time zones. And to optimise those workflows, companies need AI that truly understands how their business runs—not just what the data says, but why it matters.”

India has established itself as the epicentre of process excellence, largely due to the strength of its GCCs. These centres now serve over 150 global customers, including leading names such as Wells Fargo, Hitachi Energy, Reckitt, Alstom, Dell, Saint-Gobain, and Robert Bosch GmbH, among others.

Celonis’ innovation hub in Bengaluru, known as Celonis Garage, is at the forefront of supporting these GCCs. This hub is dedicated to accelerating AI-driven process optimisation for GCCs. By integrating Celonis’ process intelligence platform with AI, GCCs can swiftly identify inefficiencies, automate complex workflows, and optimise end-to-end business processes, enabling companies to stay competitive globally.

The synergy between AI and PI is rapidly gaining traction across industries within the GCC landscape.

“India is our future. And we think Celonis and India are like a match made in heaven,” Rinke concluded, hinting at significant investments planned in India in the upcoming months.

Published Updated

Corrections: AIM corrects errors of fact and notes the change on the story. Spotted one? Tell the editors. Read our editorial standards.

More in this story

AI · 5 stories
  1. You are here: Most AI Investments Will Fail, Unless…
  2. Can You Chat With Me After I Die? Meta Thinks So
  3. Enterprise AI Has a Massive Problem, and It’s Not the Models
  4. NeuroDx Previews Sphota, a Language Model Architecture Designed to Reduce Memory Needs
  5. India’s GCCs to Add 1.4 Million Jobs, Drive 170 Million Sq. Ft. Office Demand by 2031
About the author

Senior Tech Journalist

Shalini is a senior tech journalist, exploring the latest advancements in AI and GCC sector. When she's not reporting on the latest innovations, you can find her immersed in her next literary adventure.

All stories
Belamy · Daily

The day’s AI news, before your first meeting.