The landscape of enterprise software development reached a significant inflection point this week as Matt Calkins, CEO of Appian, addressed a global audience of developers, partners, and stakeholders. In a keynote that challenged the very category his company helped pioneer, Calkins argued that the traditional definition of "low-code"—centered on visual drag-and-drop interfaces—is rapidly becoming secondary to a more profound requirement: the underlying infrastructure of deterministic process and governance. As generative artificial intelligence (AI) increasingly handles the "building" phase of software through natural language prompts, Calkins posits that the true value of an enterprise platform now resides in its ability to act as a reliable "harness" for otherwise unpredictable probabilistic models.
This strategic pivot marks a departure from the industry’s decade-long focus on democratizing development through visual modeling. Instead, Appian is repositioning itself as a critical layer of "East Coast AI"—a philosophy rooted in reliability, structured governance, and the high-stakes requirements of regulated industries. Calkins’ thesis is clear: while AI can now generate the code or the interface, it cannot, on its own, provide the "five nines" reliability (99.999% uptime and accuracy) that global banks, pharmaceutical giants, and government agencies demand for their core operations.
The Evolution of the Low-Code Paradigm
To understand the magnitude of this shift, one must look at the historical trajectory of Appian and the broader low-code market. Founded in 1999, Appian spent its first decade focused on Business Process Management (BPM). It eventually transitioned into the leader of the "low-code" movement, a term popularized by analysts like Forrester in 2014 to describe platforms that allow for rapid application delivery with minimal hand-coding.
For years, the "low-code" identity was synonymous with the visual interface—the ability to move blocks around a screen to create a workflow. However, Calkins now suggests that these interfaces were merely "contingent tools" rather than the heart of the product. The rise of Large Language Models (LLMs) has made it possible for users to describe an application in plain English and have the system generate the necessary components. In this new reality, the "building" interface loses its primacy.
"The building interfaces are much less important now," Calkins noted during his address. "But they were never the heart of the product. Our strength is most of all in runtime power and reliability—and not merely how fast you can create the application." This distinction is vital for the company’s future valuation. If the value of a company is tied to its drag-and-drop UI, it risks obsolescence as AI matures. If its value is tied to the secure, integrated, and scalable environment where those applications run, then AI becomes a tailwind rather than a threat.
East Coast AI vs. West Coast Idealism
Central to Calkins’ vision is a distinction he draws between the "West Coast" and "East Coast" approaches to artificial intelligence. In his framing, West Coast AI—driven by Silicon Valley giants—is characterized by its pursuit of the "marvelous" and its comfort with the probabilistic nature of LLMs. These models operate on guesses; they predict the next most likely token in a sequence. While this leads to impressive creativity and conversational ability, it is fundamentally incompatible with tasks requiring absolute precision.
"AI is empirically probabilistic," Calkins explained. "Everything it says—really everything—is a guess. If you ask it what one plus one is, it will guess. That’s just the nature of the technology."
In contrast, "East Coast AI" is described as hard-nosed, pragmatic, and focused on strategic operations. It recognizes that in a business context, a "guess" can be a liability. For example, a pharmaceutical company managing a clinical trial or a bank processing a multi-billion dollar wire transfer cannot rely on a probabilistic guess. They require deterministic outcomes—actions that are repeatable, auditable, and guaranteed to follow specific rules.
Appian’s strategy is to provide the deterministic "harness" that surrounds the probabilistic "engine" of the AI. By placing AI within a structured process model, the platform ensures that the AI’s output is regulated, governed, and restricted to safe parameters.
The Two Markets of the AI Era
Calkins identifies two distinct markets emerging in the AI sector. The first is the market for the Large Language Models themselves—a space currently dominated by OpenAI, Google, and Anthropic. The second is the market for the "framework" or "structure" within which these models operate.
Appian has no intention of competing in the first market. It does not seek to build the world’s largest LLM. Instead, it aims to own the second market. Calkins argues that the more powerful AI models become, the more desperate enterprises will be for a way to control them. This "harnessing" market includes several critical components:
- Process Orchestration: Ensuring AI steps occur in the right order within a wider human-in-the-loop workflow.
- Data Fabric: Providing AI with secure, unified access to enterprise data without the need for complex data migration.
- Governance and Permissions: Restricting what an AI can see and do based on established corporate hierarchies and security protocols.
- Auditability: Creating a permanent record of every decision made or influenced by an AI for regulatory compliance.
Supporting Data: The Enterprise AI Implementation Gap
Calkins’ pivot is supported by broader industry trends. According to 2025 data from Gartner, while over 80% of enterprises have explored generative AI pilots, fewer than 20% have moved these projects into full-scale production. The primary hurdles cited by Chief Information Officers (CIOs) are concerns over "hallucinations" (probabilistic errors), data privacy, and a lack of clear governance structures.
Furthermore, a study by IDC suggests that the "AI Orchestration" market is expected to grow at a Compound Annual Growth Rate (CAGR) of 35% through 2030, outpacing the growth of the underlying model market. This suggests that the "East Coast" demand for reliability is a significant and underserved segment of the economy. Appian’s existing infrastructure—which already handles millions of critical transactions daily for organizations like the U.S. Air Force and Merck—positions it to capture this demand.
Chronology of Appian’s Strategic Shifts
- 1999–2010: The BPM Era. Appian focuses on Business Process Management, helping companies automate complex back-office workflows.
- 2011–2022: The Low-Code Expansion. The company pivots to "Low-Code," emphasizing its visual interface to speed up app development. It goes public in 2017.
- 2023–2024: The AI Integration Phase. Appian introduces "AI Skills" and "Data Fabric," beginning the process of weaving AI into its existing process automation stack.
- 2025–2026: The Infrastructure Pivot. At Appian World 2026, Calkins declares the "building" interface secondary, focusing the company’s mission on providing the deterministic harness for the AI-driven enterprise.
Official Responses and Industry Implications
Industry analysts have reacted with cautious optimism to Calkins’ transparent admission that low-code tools are being superseded by AI. "It is rare to see a CEO acknowledge the potential obsolescence of their primary category-defining feature," said one senior analyst at Forrester. "However, by moving the goalposts from ‘building’ to ‘running,’ Appian is attempting to insulate itself from the commoditization of code generation."
Competitors in the space, such as Microsoft (with Power Platform) and ServiceNow, have also been integrating AI heavily, but their messaging often remains focused on the "co-pilot" aspect—helping users build faster. Appian’s focus is notably different; it is less about the speed of creation and more about the safety of the execution.
For regulated industries, this message is particularly resonant. In sectors where "move fast and break things" is a recipe for legal and financial ruin, the promise of an AI vehicle that comes equipped with "brakes and a steering wheel" (the deterministic harness) is a compelling value proposition.
Impact on the Future of Work and Software
As Appian leans into its role as an AI infrastructure provider, the implications for the workforce are significant. The "citizen developer" movement—once the darling of the low-code world—may shift. If natural language is the new interface, then the skill of the future is not learning to use a drag-and-drop tool, but rather understanding how to define a process and its constraints.
Calkins remains unsentimental about the legacy applications that AI might replace. "I don’t sympathize with the legacy applications," he stated. "We’re on the side of the replacement." By positioning Appian as the foundation for these replacements, Calkins is betting that the company will not just survive the AI revolution but become the essential scaffolding upon which the modern, AI-integrated enterprise is built.
Ultimately, the success of this strategy depends on whether enterprises view the "harness" as a distinct layer of the tech stack or if they expect model providers to eventually solve the reliability problem themselves. For now, Calkins is betting on the inherent nature of the technology: that probability will always require a partner in determinism. In the high-stakes world of enterprise operations, "good enough" is rarely enough, and Appian’s gamble on "five nines" reliability may well define its next decade of growth.
