Artificial intelligence has undoubtedly dominated corporate technology conversations since the debut of ChatGPT four years ago, thrusting IT departments into the center of organizational strategy. Yet, despite trillions of dollars in global investment and relentless hype, many business leaders are confronting a peculiar paradox: the productivity gains delivered by AI have frequently fallen short of expectations. According to Appian CEO Matt Calkins, speaking at the Appian Europe conference, this disconnect stems from the fundamental nature of the technology itself. Characterizing AI as inherently ‘weird,’ Calkins pointed out that unlike past technological revolutions that began with users who had the most acute business needs, the current AI boom started with lower-value use cases focused primarily on personal productivity and minor cost savings rather than core enterprise operations.
The root of this challenge lies in the tension between deterministic business software and probabilistic artificial intelligence. Large organizations—such as banks, pharmaceutical companies, and public-sector institutions—operate at immense scale and maintain the least tolerance for error. These enterprises have spent decades standardizing and governing complex workflows to ensure repeatability and reliability. Consequently, they remain hesitant to deploy unfiltered, unpredictable AI models directly into high-volume, mission-critical processes where failure could introduce unacceptable operational chaos. Calkins argues that for artificial intelligence to deliver genuine value within the core of the economy, it must be embedded within mature, structured control frameworks rather than left to operate as a standalone, unconstrained technology.
To bridge this gap and make AI reliable enough for serious enterprise work, Calkins outlined four core capabilities that organizations must wrap around AI models: deterministic process orchestration to keep them bounded, a distributed data fabric to keep them grounded in corporate data, robust governance infrastructure to monitor and constrain behavior during execution, and the flexibility to route tasks to the most appropriate execution option—whether that is a frontier model, a small language model, a robotic process automation bot, or a human worker. These foundational elements closely mirror Appian’s established platform strategy, which combines process automation, data management, and governance capabilities into a unified operational environment.
Marking a significant evolution from theoretical discussion to commercial execution, Appian used the European event to announce the launch of a new enterprise control plane in beta. Describing the control plane concept, Calkins emphasized that observation serves as the essential baseline. Organizations need visibility into what software agents are doing, how much they cost, and whether they are complying with established boundaries. Beyond basic monitoring, however, Appian’s control plane integrates autonomous agents directly into structured business process orchestration, ensuring they operate within predefined logical sequences.
This architectural decision—keeping coordinating authority outside the agents themselves—addresses a major risk in autonomous systems. By enforcing a rule where agents interact with a deterministic layer rather than calling other agents directly, enterprises can delegate specific tasks to AI without relinquishing control over wider business operations. The process itself continues to dictate how work unfolds, where human intervention or exceptions are managed, and when traditional automation is more suitable.
As enterprises navigate the next phase of digital transformation, the introduction of dedicated control planes highlights a growing industry emphasis on governance and operational reliability. By anchoring probabilistic artificial intelligence within established deterministic workflows, software providers aim to move past the initial wave of hype and deliver measurable, scalable business outcomes for organizations operating in high-stakes environments.