Cloudera forecasts disposable apps & AI governance shift

Cloudera executives expect 2026 to bring sharp shifts in how organisations build, govern and store data for artificial intelligence, including short‑lived AI-generated applications, stricter oversight of "Frankenstein" systems and pressure on data storage capacity.

Their forecasts point to a period of rapid experimentation. They also highlight rising concern over costs, controls and infrastructure limits as AI moves deeper into day‑to‑day operations.

Chris Royles, Field CTO for EMEA at Cloudera, said AI will reshape how software is conceived and delivered.

"In 2026, AI will start to radically change the way we think about apps, how they function and how they're built. Today, apps are declarative, with millions of lines of code coming together to follow fixed rules. But AI is tearing up the rulebook and removing these constraints. With just a few lines of code and a prompt, users will be able to request temporary, purpose-built modules that replace the need for dedicated apps, with AI essentially acting as the operating system and app developer.

"For example, a user could ask AI to coordinate a call between team members. Once that function has served its purpose, it closes. In the background, AI then quietly refines the module based on its own evaluation of the experience and feedback until the same function is needed again. These 'disposable' apps will constantly evolve through self-learning and can be built and rebuilt in seconds.

"Strong security and governance will be critical in this application development model, as organisations will need visibility into the reasoning processes AI uses to create and evolve each module. This will ensure errors are corrected and improvements are made safely and transparently."

Royles expects this model to alter the role of traditional, long‑lived applications. It may also change how organisations think about software lifecycles and maintenance.

Patchwork systems

Cloudera executives also warn that early‑stage AI builds may run into financial and compliance barriers. Many organisations have combined different AI components in single services.

Paul Mackay, RVP Cloud EMEA & APAC at Cloudera, said some of those systems are already under review.

"Many organisations will begin shelving their 'Frankenstein' AI applications they built for specific business use cases - as costs spiral and governance concerns grow. As organisations start to deploy AI tools and applications, the risks of stitching together agentic AI, large language models, low- and no-code platforms, and open-source components in the name of innovation become amplified. The result will be a patchwork system, whether it's an AI helpdesk cobbled together from a no-code form, an LLM auto-responder, or an agent attempting fixes in the background. But these patchwork systems could become money pits, tangled up in unpredictable token usage and growing compute demands. They are also proving increasingly difficult to govern effectively.

"As observability and governance take centre stage, organisations will reassess whether these applications need to be rebuilt or redesigned altogether. By maintaining visibility into the data that feeds these systems during development or restructuring, organisations can regain control over compliance and costs. Without this, many ambitious AI projects risk becoming too costly to sustain."

Cloudera expects this reassessment to drive more structured approaches to AI design. It may also push spending away from experimental projects that lack clear controls.

Autonomous oversight

Alongside concerns over ad hoc builds, the company anticipates new roles for AI in governance itself. These would sit inside existing data operations.

"In 2026, we will see the emergence of specialist AI agents dedicated to data governance. These digital colleagues will continuously monitor, classify, and secure data wherever it resides, ensuring governance becomes an always-on function embedded into daily operations," said Wim Stoop, Senior Director, Cloudera.

"For example, a compliance agent could flag risks across business systems, while a security agent automatically adjusts access permissions, as new data enters the environment, all without human intervention. As these digital colleagues become part of the organisational fabric, businesses will need to manage them much like human teams - tracking skills, performance, and collaboration through new frameworks for 'agent resource management.

"Governance will no longer be something people do, but something they oversee. Rather than manually enforcing every rule, humans will instead 'govern the governance' - shaping the process as it runs. As trust in these systems grows, we'll be able to step back and let continuous governance take over, delivering cleaner data, stronger compliance, and AI-ready insight that drives real business value."

Stoop said this shift would create new management structures around AI agents. It would also demand new tools for tracking their behaviour.

Data limits

Cloudera's outlook also includes pressure on data storage. The firm sees a change in how organisations treat human‑generated and synthetic information.

"As global data storage nears breaking point, human-generated data will rise sharply in value. The era of digital hoarding - keeping everything simply because storage was cheap - will come to an end, forcing organisations to decide what's worth keeping and what needs to be generated again. AI-generated data will become disposable, created and refreshed on demand rather than stored indefinitely.

"This shift will redefine how data is valued. Businesses will compete to source and protect authentic, human-created data to train and differentiate their AI models. This shift will spark a new data economy - one that prizes originality, context, and quality over sheer volume, redefining what makes information valuable in the age of synthetic content."

Executives at Cloudera expect these trends to shape investment decisions in AI and data infrastructure over the next year as companies test new applications and weigh the long‑term costs of their data strategies.