Enterprise AI Governance: From Features to Financial Exposure
Enterprise AI governance is the set of policies, controls and financial rules that decide how AI systems are built, funded, monitored and adapted, ensuring that powerful models become safe, auditable software rather than uncontrolled experiments that quietly reshape processes and costs without clear ownership or accountability. The uncomfortable truth is that AI capability is no longer the hardest part of enterprise software. Models can already generate working applications, integrations and workflows at a pace traditional IT could never match. The real fight now is over who controls that power, how it is funded and how its risks are contained. According to Deloitte’s China Widener, the next phase of enterprise AI "will be defined less by technology and more by governance, finance and strategy". Organizations that treat AI like another license purchase will discover they have bought a new balance sheet, not a tool.
Shadow IT Risk: AI-Built Apps Without Owners
AI is supercharging shadow IT risk. A business user can now describe an application in plain language and see a usable first version take shape within minutes. That speed feels liberating to teams tired of waiting for IT, but it also means "a fully functioning application can now be created before IT has a clear view into what it does, what data it reaches, or who will support it". Shadow IT has always meant tools outside formal oversight; shadow AI adds opaque logic, hidden data flows and unpriced compute consumption. IBM’s 2025 Cost of a Data Breach Report found that one in five breached organizations tied incidents to shadow AI, and that high levels of shadow AI added USD 670,000 (approx. RM3,081,000) to average breach costs. The build problem is solved; the ownership problem is wide open—and now it carries direct operational and financial consequences.
Low-Code AI Platforms Grow Up: Deterministic Code and IT Controls
The low-code/no-code story is being rewritten around deterministic code execution and enterprise AI governance. After eight years of building a no-code automation platform, Tines launched 3B on Tuesday, a new platform that uses AI to author enterprise workflows but still uses conventional code to execute them. Large language models became good enough at writing code that the visual builder itself looked dated. With 3B, employees in finance or HR can describe the application or automation they need, while IT decides upfront which systems and data they can touch. Once generated, each workflow step becomes code that runs in its own ephemeral Docker container, executing as a deterministic workflow instead of relying on a model to reason through every step at runtime. That design choice is opinionated: building is commoditized; reliability comes from controlled, inspectable code and clear access maps, not from clever prompts.
AI Economics: Activity-Based Costs Force Strategy to the Fore
Enterprise AI is blowing up the old software finance model. For decades, executives bought platforms, finance approved budgets, and costs scaled predictably with seat counts. Now AI agents, token-based pricing and new infrastructure models mean organizations increasingly pay for activity—AI inference, API calls, workloads, data access and computation—rather than users. Those nuances are reshaping executive responsibilities, because infrastructure choices once buried inside IT now carry direct financial implications. Widener argues that organizational structure—not technology—is one of the biggest barriers to successful AI adoption. AI strategy has become a system decision: contracts, architecture, governance and financial management must be designed together, or enterprises will end up with software that works but is misaligned with their risk appetite and budget. Looking ahead, she expects hybrid architectures that balance vendor capabilities with more internal flexibility, not all-in bets on a single stack.
Specification, Governance and the Next Competitive Edge
The next competitive advantage in enterprise software will come from strategy and governance frameworks, not raw AI capability. Gartner predicts that task-specific AI agents will appear in 40 percent of enterprise applications by the end of 2026, up from less than 5 percent in 2025. When AI agents are everywhere, being able to build one is no longer a differentiator; being able to govern hundreds is. Many AI-assisted development tools still treat generated code as the main output, but for enterprises the real asset is the specification where business intent lives. A structured specification "gives governance something concrete to attach to" and lets ownership, compliance and change management latch onto AI-built apps instead of chasing them. Vendors that design low-code AI platforms with deterministic execution, permissions, and access maps show the direction of travel: governance has to be part of how software gets built, not an audit trail added later. The organizations that win will be those that treat enterprise AI governance as a first-class product feature and a board-level discipline.
