The Boston SoftDesign team attended the 2025 World of Workato conference in Las Vegas, NV, in August 2025. Our main takeaway was this: AI is a powerful tool with huge potential, but we need to be realistic about where AI is at now—and how organizations can actually benefit from using it.

Wherever you look, AI—or, at the very least, marketing for AI—seems to be there. Venture-backed platforms and enterprise vendors alike are making bold claims about AI-powered productivity gains and predicting that artificial general intelligence is just around the corner. Yet, many of these platforms’ solutions remain unproven in production—and some never had autonomous capabilities at all, which has spurred accusations of “agent washing.” 

Interestingly, Gartner forecasts that more than 40% of agentic AI projects—those designed to act autonomously—will be canceled by the end of 2027, often due to unclear business value, rising costs, and inadequate governance frameworks. This statistic underscores a critical point: companies should avoid falling into the hype-driven trap of enterprise-wide AI deployment. Instead, a focused, incremental approach offers a more reliable path to meaningful outcomes.

The Hype and the Reality of Agentic AI in 2025

Many vendors are positioning their AI solutions as breakthrough tools capable of independent task execution, only to deliver limited functionality. This “agent washing” rebrands legacy technologies, such as chatbots or RPA, as autonomous agents without the capabilities that a truly autonomous solution would possess, such as end-to-end customer support or system monitoring with real-time proactive maintenance recommendations. Additional analysis suggests that early AI initiatives often falter during pilot stages, with only a small share transitioning into sustainable production.

Why an Iterative AI Strategy Makes More Sense

A targeted project-by-project approach offers several advantages:

  • Controlled scope and measurable results. Teams need to track AI agent interactions with a tool like Workato’s Acumen, which can gather user conversation metrics and offer real-time alerts and recommendations. This close monitoring enables gradual refinement and helps prevent unwanted drift in bot behavior.
  • Real-world stability before expansion. Rather than going all-in on enterprise-wide agentic AI deployment, starting small—say, with customer support SLAs or transaction verification—allows teams to understand nuances and fine-tune responses daily, building a foundation and confidence in the results of their AI tools.
  • Precision through process automation. While AI can orchestrate complex logic with remarkable capability, the foundation of reliable operations lies in robust workflow and process automation. Automated processes deliver the precision and consistency that sectors like finance demand, where even minor deviations can have significant consequences. The most effective approach leverages automation for reliable, repeatable tasks while ingesting AI where it truly excels, maximizing both reliability and innovation without over-dependence on either technology.

What’s at Stake in a Failed AI Deployment

Organizations pursuing enterprise-wide AI often overestimate the maturity of the technology and underestimate the risks to their operations and data security. This misalignment can cause projects to stall, incur budget overruns, or simply be discontinued. As analysts at Boston Consulting Group note, relying on flashy models without aligning with operational workflows or investing in change management are common errors.

By contrast, focusing on specific outcomes—reducing response times, improving service quality, delivering predictable SLAs—can deliver immediate, tangible benefits and create a better foundation for expansion.

Boston SoftDesign’s Outlook on AI Deployment

Boston SoftDesign’s team of experts regularly reinforces the importance of strategic, well-governed digital transformation. For instance, our writing on system integrations emphasizes that success lies in structured ecosystems—not in deploying isolated technologies without governance. Similarly, forward-looking reflections on AI highlight the need to align innovation with business value over flashy tools that impress stakeholders but don’t deliver for the teams using them.

AI’s long-term promise lies in its cumulative, measured application, not in sweeping, enterprise-wide rollouts. With Gartner’s forecast that over 40% of agentic projects may be canceled by 2027 as a backdrop, the prudent path forward starts with pilot projects that deliver real value, backed by proper monitoring, iteration, and governance.

By building incrementally, refining continuously, and scaling only when outcomes are solidly proven, organizations stand a far better chance of transforming enthusiasm into sustainable performance, rather than fleeting hype.