For technology professionals, artificial intelligence is no longer a novelty—it is the defining force shaping the sector’s future. Yet as enthusiasm builds around AI pilots and prototypes, a stark truth is emerging: without robust operational foundations, AI is unlikely to deliver lasting value.
Speaking at the Gartner IT Symposium 2025, Dave Stevens, founder and Managing Director of Brennan, warned that success with AI depends not on hype but on what he calls “Operational Innovation.” This approach centres on strategy, data quality, governance and culture—the often overlooked foundations that turn AI from experimental to essential.
The need for this shift is pressing. Gartner research cited by Stevens predicts 60% of AI projects will be abandoned by the end of 2026 due to inadequate data infrastructure. At least 30% of generative AI initiatives are expected to fail after proof-of-concept, while more than 40% of agentic AI efforts may collapse by 2027, hindered by rising costs and weak business outcomes.
This wave of project abandonment, said Stevens, reflects the dangers of “dream selling”—promoting AI as a cure-all without embedding it in practical, outcome-focused planning.
To counter this, he outlined four pillars necessary for sustainable AI adoption: Strategy: Stevens stressed the need for clear goals. Around 75% of digital transformations fall short, and just 5% are completed on time and within budget. Asking the right questions upfront—what problem is being solved, what benefits are expected, how will success be measured—is key to avoiding expensive missteps. Data and Identity: AI systems are only as reliable as the data they ingest. With 65% of organisations lacking effective data management, hallucinations—false but confident outputs—are common. Stevens recommended role-based access, metadata tagging and domain-specific data libraries. In the utilities sector, such practices helped cut chatbot escalation rates by improving answer accuracy. Governance, Risk and Compliance: Effective AI at scale requires embedded controls. Stevens cited a mining client that used AI to automate contract review, significantly reducing errors by layering access restrictions, policy checks and quality controls into the workflow. Culture: The human factor is often ignored. Only 23% of organisations have formal AI training, and just 6% make it mandatory, according to ADAPT. Stevens argued that trust and storytelling are vital to reposition AI as an enabler. Brennan’s own staff embraced AI after training reframed it as a supportive “mentor.” Compounding these barriers is the burden of legacy systems. Gartner reports that 62% of tech leaders believe outdated models obstruct both current and future strategic goals, making simplification essential before layering AI solutions.
The lesson, said Stevens, is simple: organisations must first “till the soil” before expecting results. Operational Innovation is not optional—it is the groundwork for AI to evolve from promise to performance.
As the global AI conversation matures, this practical focus offers a route for the UK and others to lead in delivering responsible, effective AI at scale. Only by coupling ambition with operational discipline can AI fulfil its transformative potential.
Created by Amplify: AI-augmented, human-curated content.