AI tools have quickly become a standard part of everyday workflows and business functions. Yet, despite widespread adoption, many organizations are running into the same constraint: AI is generating insights but not necessarily driving measurable business outcomes.

Research from McKinsey shows that while 88% of organizations are using AI in at least one business function, only around one-third have scaled it across the enterprise, and just 39% report measurable financial impact, highlighting the gap between individual adoption and team or organization-wide execution that drives results.

Most deployments remain in what could be described as the “chat phase” where AI is in the hands of every person, summarizing information and answering questions. These tools are undeniably helpful, reducing time spent searching for information while accelerating knowledge work and lowering the barrier to entry for interacting with data.

However, there is a disconnect with these AI conversations often existing outside the systems where work is actually planned, tracked, and delivered. An AI assistant might identify a project risk, summarize delivery status or suggest next steps, but the act of turning that insight into action still requires human intervention.

There remains a heavy reliance on coordination across tools to manually update plans, assign tasks or trigger workflows. Execution remains fragmented and human-dependent. It also caps AI’s impact at the individual level. While people may feel more productive, the organization as a whole doesn’t necessarily move faster of make better decisions.

This gap is quickly becoming one of the defining challenges in scaling AI beyond individual productivity to team‑level impact and, ultimately, organizational transformation.

Why context is critical

A key barrier to progress is that most AI systems still lack access to meaningful operational context. While models are highly capable, they often operate without visibility into the structured data that defines day-to-day work, project plans, dependencies, risks, and resources. Without a trusted, governed data foundation, the risk of AI projects failing increases.

And when only half (54%) of AI projects make it from pilot to production, this underscores how difficult it is for organizations to move from experimentation to scaled, operational impact.

Without this context, AI can generate recommendations but cannot easily support execution. Teams are left to interpret outputs and translate them into action across multiple tools. This manual translation slows things down and introduces inconsistency and risk.

Successful AI adoption that moves beyond the chat phase requires enterprises to invest in trusted, governed access to systems that can manage real-world complexities like messy data and cross-functional workflows. When AI and business teams can see the same data and interact with it securely, enterprises can move from generating insight to driving execution.

One example of this approach is the emergence of open architecture that allows AI tools to connect cleanly to enterprise data and workflows. We recently launched our Model Context Protocol (MCP) Server with the aim to address this gap by securely connecting AI tools directly to live work data. This means teams can connect their own preferred tools and embed AI into the systems where work actually happens.

Early usage illustrates how this enables a shift from insight to action: 48% of all actions taken by early adopters using Claude via our MCP Server actively move work forward by automatically creating tasks or making updates, rather than simply requesting information. By operating this way, organizations can move from isolated insights to AI as the intelligence layer connecting every function and system. When AI works across an organization this way, the gains compound.

Measuring AI ROI

As AI adoption matures and scales across enterprises, organizations are under increasing pressure to demonstrate clear return on investment, not just model performance or usage metrics.

The focus is shifting toward how AI drives measurable business outcomes. Instead of focusing on AI adoption at the individual level, leading organizations are measuring impact at the team and organization level. They ask different questions: Does AI reduce cycle times? Does it surface risks earlier? Does it improve decision quality across teams?

At the same time, enterprise-wide adoption is bringing the cost of AI into sharper focus. Decision-makers are looking beyond experimentation to understand how to scale AI effectively and sustainably. This is driving greater emphasis on cost predictability, as well as more efficient models of AI consumption that optimize how data is shared and processed.

Together, these priorities are reshaping how organizations evaluate AI, balancing performance with disciplined spend to ensure long-term, scalable impact.

A more operational future for AI

Taken together, these trends point to a broader shift in enterprise AI. The future is not about adding more standalone tools, rather, it is about positioning AI as the intelligence layer connecting every person, function and system within an enterprise.

The next phase of adoption will be defined less by model capability alone and more by how effectively AI is embedded into the fabric of day-to-day work. That means connecting insight to execution, while maintaining the governance and context required to operate at scale.

It is also in this next phase that competitive advantage will not come from access to AI alone, but from how effectively organizations connect it to their operational systems, data, and governance frameworks.

As the industry moves beyond the “chat phase,” the organizations that succeed will be those that treat AI not just as a productivity tool, but as part of their operational infrastructure, balancing autonomy with control, and innovation with accountability to drive measurable outcomes.

We feature the best data visualization tools.

This article was produced as part of TechRadar Pro Perspectives, our channel to feature the best and brightest minds in the technology industry today.

The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/pro/perspectives-how-to-submit