Business leaders face mounting pressure to handle unforeseen operational costs, vendor "agent washing" and increasingly complex data governance requirements as they seek measurable value from artificial intelligence (AI) investments, says global tech research firm Gartner Inc.
Organisations should establish AI adoption roadmaps, clear governance frameworks and disciplined risk controls that support sustainable returns over the long term.
Speaking at the recent Gartner Data & Analytics Summit, Jorg Heizenberg, vice-president analyst at Gartner, noted a stark divergence in cost perceptions surrounding AI deployment.
While 60% of IT leaders express concern that AI agents could drive unexpected financial burdens, only 20% of data and AI leaders are concerned unpredictable costs might limit AI value.
Georgia O'Callaghan, director analyst at Gartner, said measuring AI success requires moving past traditional financial metrics.
She highlighted three strategic pillars for enterprise value: return on intelligence, achieved by setting an organisational AI ambition; return on integrity, driven by robust data foundations; and return on individuals, centred on human-led transformation.
"Without strong foundations, AI is just amplified ignorance. It isn't about humans in the loop, but about humans in the lead," said Ms O'Callaghan.
This involves collaborating with stakeholders to set a clear AI ambition based on the organisation's tolerance for disruption, whether that means taking a cautious approach, an opportunistic one, or daring to be an AI-first pioneer.
She warned that complex vendor pricing models, which rely on metrics such as graphics processing unit hours and token consumption, can easily lead to unexpected costs.
"In some cases, the cost of tokens used for AI coding assistants has already surpassed the cost of actual software developers," said Ms O'Callaghan.
If they compare complex AI workflows without proper oversight, organisations might learn "mostly from the bill", she warned.
Enterprises are advised to implement "cost-driven design" to evaluate financial impacts early during the prototyping phase, according to Gartner, while tying AI investments to tangible outcomes, both financial and non-financial, such as improved customer experiences.
Agent Washing
As companies increasingly eye autonomous systems, Pieter den Hamer, vice-president analyst at Gartner, warned that the current AI agent market is highly volatile and heavily marketed.
"Agentic AI promises transformation, but today's reality is shaped by hype, limited integration maturity and emerging lifecycle management capabilities," said Mr den Hamer.
While current large language model-based agents effectively perceive information, they still fall short in terms of adaptability and contextualised decision-making.
Furthermore, the rapid surge in interest has prompted some tech vendors to engage in agent washing: rebranding existing, non-autonomous legacy capabilities to capitalise on market demand.
He recommended organisations initially prioritise low-complexity use cases and aggressively train their workforce.
The surge in generative AI (GenAI) deployment is also forcing a massive overhaul in enterprise data management frameworks.
Guido De Simoni, vice-president analyst at Gartner, said governance models must evolve to handle exponential growth in metadata and unstructured data. However, the transition may not require completely replacing existing setups.
Gartner forecasts 80% of GenAI business applications will be developed directly on existing data management platforms. This integration approach is expected to reduce deployment complexity and cut time-to-delivery by half.
