DEVOPSdigest

AI investment is growing 52% year-over-year, yet progress is bumpy given data challenges and a skills gap that is holding businesses back. While developers are excited about the next model release, the reality is 99% of enterprises face AI project disruptions that are often not related to model choices that organizations often stress over. AI innovation hype is outpacing enterprise readiness, leaving developers with a tough choice: move fast on AI and risk failure, or move cautiously and watch competitors pull ahead. But what if there's a third option — one that requires no compromise?

Couchbase's survey of 800 IT leaders found that strategic experimentation backed by a deep understanding of the organization's data lets them capture AI's benefits and help mitigate financial losses. According to the report, businesses unable to effectively use AI in a timely manner could lose on average 8.6% of their revenue per month. Based on the research sample size, that's an average annual loss of almost $87 million per company.

As it turns out, organizations that encourage more experimentation from their developers see a higher proportion of AI projects enter production than those that have strict parameters on experimentation.

Experimentation as a Path to AI Success

AI innovation is uncertain. New models, protocols and approaches to AI continue to emerge, making it difficult for organizations to adapt and keep pace. While jumping in blindly risks failure, waiting too long risks losing a competitive edge.

Strategic experimentation can provide the balanced approach organizations need to navigate this challenge. The organizations succeeding with AI see "failure" as research and development, not waste. When developers run controlled pilots, teams can validate AI use cases and better understand what does and doesn't work before scaling.

The research shows the payoff:

■ Companies encouraging experimentation see 10% more of their AI projects reach production. They also waste 13% less on AI investment than their more restrictive peers.

■ 81% of CIOs agree that education and experimentation are critical for AI development, and 74% say even failed projects have value because they provide insights for next time.

The AI-ready companies are those that manage risk through structured experimentation, using developer-friendly techniques and practical guidelines such as:

■ Running low-risk proofs of concept in sandbox environments.

■ Using A/B testing and/or feature flags for AI rollout.

■ Allocating a "failure budget" to encourage responsible risk-taking.

■ Building cross-functional teams of developers, data engineers and product managers to pilot new models
.
■ Investing in education and developer tools that are familiar and easy to use.

For experimentation to pay off, however, it must be paired with strong data practices that keep AI models grounded in reality.

Successful AI Starts With Robust Data Practices

Even the best AI strategy can't overcome bad data. Poor quality inputs lead to AI agent drift, hallucinations and/or biased models that damage trust. Equally important is timeliness. If data isn't recorded and accessible in real time, AI will base decisions on outdated information.

Another hazard: Governance gaps create hidden risks. Without clear access controls and monitoring, AI systems can expose sensitive data or violate compliance requirements. That's why 79% of CIOs say they need strict governance in place to succeed with AI. However, as AI architectures grow more complex, they become harder to govern, limiting control and increasing enterprise risk. In fact, 75% of CIOs believe now is a great opportunity to consolidate and simplify technology stacks.

Teams also face major knowledge and experience gaps: 62% of enterprises do not fully understand where they are at risk from AI (e.g., through security or data management issues). Another 72% of enterprises say their level of data understanding and control needs to be higher than before to use AI effectively and safely. Complementing this, 25% of CIOs cited that a lack of skills has prevented them from delivering AI projects, while 21% said it has disrupted ongoing AI projects.

Here's the upside: Organizations with greater understanding of their data have more control and are 33% more likely to be ready for agentic AI, giving them an edge in automation and advanced use cases. There are several practical steps developers can take to better understand their data and achieve AI success:

■ Map where AI-relevant data lives and how it flows through pipelines.

■ Build observability into data processes to catch quality or latency issues early.

■ Consolidate fragmented systems to reduce silos, complexity and risk.

■ Use flexible data storage (like JSON) to accommodate the fluid nature of data that AI generates and consumes.

Closing the Gaps Between Vision and Execution

Many organizations silo experimentation and data control as if they are unrelated, but in reality, they reinforce each other. With data fragmentization, experiments can produce misleading results. Without a safe environment for experimentation, teams hesitate to deploy AI projects, even when they have the technical ability. This combination leads to delays, with 17% of CIOs reporting their AI budgets are going to waste and goals are delayed by six months or more.

Since organizations on average report their current setups have a lifespan of just 18 months before they can no longer support AI applications, it's important for them to build the right foundation for AI. That means putting clear data controls in place and using an architecture that can support multiple AI needs. Companies seeing success share common strategies:

■ Using a platform that supports the management of all data types involved in AI interactions.

■ Consolidating data systems to handle operational, analytical, AI and mobile workloads on a unified platform.

■ Enabling real-time data access, so models can work with current information.

■ Supporting features like vector search and multi-purpose data handling.

Turning AI Investment Into Competitive Advantage

The path to AI success is all about building the right foundation that enables both confident experimentation and rapid scaling. This starts with simplifying data architectures. Indeed, all surveyed enterprises are consolidating their AI tech stacks because fragmented systems create barriers to effective AI deployment. Teams with these data practices move faster because they can iterate without fear of catastrophic failures.

Investment patterns unveiled in the research show agentic AI, GenAI and traditional AI/ML approaches are receiving nearly equal funding despite their different maturity levels. With 78% believing early adopters will dominate their industries, what ultimately matters is whether your data foundation can support the iterative testing and scaling that separates successful AI initiatives from expensive experiments.