Saturday, 8 August 2026
Companies that succeed with artificial intelligence will not necessarily be those that adopt the technology first or impose the strictest controls, but those that build the organisational and technical foundations needed to support it safely, delegates heard during a keynote presentation at AI4 2026 in Las Vegas.
The presentation argued that enterprise AI has entered a new phase in which employees are increasingly deploying generative AI tools to solve business problems, often without formal approval from their organisations. Rather than viewing this trend solely as a governance issue, the speakers suggested it reflects growing demand for faster decision-making and more efficient workflows across large enterprises.
According to findings from a survey of 900 chief executives from companies with annual revenues exceeding USD 500 million across eight countries, 96% believe employees are already using generative AI without formal approval, while 42% said they believe at least half of their workforce is doing so. The survey results were presented by the speakers during the session and were not independently verified.
The keynote compared today’s AI adoption with the widespread introduction of spreadsheet software more than four decades ago. Business users rapidly embraced spreadsheets to solve operational problems before many IT departments had established governance frameworks, a pattern the speakers suggested is now repeating itself with AI agents.
Unlike spreadsheets, however, AI systems increasingly perform actions beyond analysing data. Modern AI agents can communicate with customers, automate workflows, generate reports, access corporate information and interact with enterprise systems, increasing the potential impact of errors or poorly governed deployments.
One demonstration highlighted how a single incorrect data reference in an AI-generated business dashboard could produce convincing visualisations while leading decision-makers to entirely inaccurate conclusions. The example was used to illustrate the growing importance of transparency and verification as organisations expand AI adoption.
The speakers argued that traditional IT operating models, developed when enterprise technology was scarce and highly centralised, are struggling to keep pace with the rapid adoption of AI across business units. Instead of positioning IT primarily as an approval authority, organisations were encouraged to evolve IT into a provider of secure platforms, governance, identity management, access controls and operational oversight.
A central theme of the presentation was that AI ownership should increasingly reside with the business teams responsible for operational outcomes. Under this approach, departments deploying AI agents would also become responsible for measuring productivity, return on investment and ongoing performance, in much the same way they manage human resources and operational budgets.
The keynote also highlighted the growing challenge of AI visibility inside large organisations. One example described a global manufacturer that initially believed it had approximately 40 production AI agents before a broader assessment identified more than 400 operating across different business functions, many without clear ownership or central oversight. While the example was anecdotal, it reflected a broader concern within the enterprise AI sector regarding so-called “shadow AI” deployments.
To address these challenges, the presenters called for organisations to establish enterprise-wide AI management capabilities, including visibility into which agents are operating, who owns them, what systems they can access, how much they cost and how effectively they perform.
The presentation concluded that successful AI strategies will depend less on restricting adoption and more on creating governance frameworks that allow business users to innovate while maintaining security, accountability and operational control.
As enterprise AI adoption accelerates, many organisations are shifting their focus from deciding whether employees should use AI to determining how that use can be managed responsibly, securely and at scale.
© 2026 cij.world
