You’re Not Behind: Almost Nobody Is At The AI Finish Line
Spend time on LinkedIn and you’d be forgiven for thinking that every other business has already automated half its operations. AI agents are presented as managing complex processes, while an expanding cast of “experts” claims to have transformed entire companies over a weekend.
The reality is rather different. Of course, many organizations are using tools such as Microsoft 365 Copilot to save time on individual tasks, while others are beginning to automate parts of their workflows. However, very few have redesigned how the business operates around AI – and that’s because the market is still learning how to turn access into meaningful business impact.
Jonni Niemann, Senior Product Marketing Manager, WorkPoint
Most organizations are still finding their way
McKinsey describes three broad horizons of AI transformation. The first is ‘Enablement’, where employees are given general-purpose AI tools to assist with their existing jobs. The second is ‘Automation’, where organizations begin using AI to improve cross-functional workflows at scale. The third is ‘Reinvention’, where roles, processes, and operating models are redesigned around what AI makes possible.
However, only 11% of these 750 AI users surveyed by McKinsey in the report said their organizations had reached Reinvention. Nearly half, 46%, remained at the Enablement stage, while 43% were focused on Automation.
The findings suggest that being at the first stage is not a failure. It remains a familiar position for many organizations, while even those that have made greater progress are still working out what an AI-enabled operating model should look like.
So what will help them move forward?
Access to AI is not the same as impact
Providing employees with AI tools can produce immediate benefits. Someone might write a first draft faster, summarize a long email thread, prepare for a meeting, or find a useful starting point for a piece of research.
These gains matter, but they’re often personal and difficult to translate into wider business results. As we’ve previously explored, the challenge is to move from AI innovation to business value by connecting AI to the processes and information on which the organization depends.
Say a business that buys Copilot licenses for its employees and supports the launch with training and an internal adoption campaign. Six months later, usage looks healthy and employees say they are saving time… but the finance team cannot identify a significant change in costs, revenue, delivery times, or customer outcomes.
Nothing has necessarily gone wrong per se. AI has helped people perform individual tasks within the organization’s existing way of working. But the surrounding processes, responsibilities, handovers, and approval routes have remained much the same.
This helps to explain why greater usage does not automatically produce greater value.
An employee completing a task faster is not the same as the organization improving the way that work moves from beginning to end.
This difference is visible in McKinsey’s findings on enterprise value at each stage of AI maturity. Some 48% of leaders in the Reinvention stage reported meaningful enterprise value from AI, compared with 24% in automation and just 13% in enablement.
The technology is only one part of the change
There is a useful historical parallel here. When manufacturers first replaced steam engines with electric motors, they didn’t immediately realise the technology’s full value because many kept the same factories, layouts, and working methods, replacing one power source with another.
The largest productivity gains came later, when businesses redesigned factories around what electricity made possible. Machinery no longer had to be organized around a central source of mechanical power, so production lines, equipment placement, and working practices could all be reconsidered.
Those early factory owners weren’t slow or incompetent; it’s just that the technology arrived before anyone fully understood the organizational shape needed to take advantage of it. Businesses face a similar challenge with AI. Adding a more capable tool to an existing process can make parts of that process faster, but it does not necessarily change the performance of the process as a whole.
According to McKinsey, leaders were 5.3 times more likely to report enterprise value when workflows had been redesigned than when they remained unchanged. Yet “redesign the workflow” is easier to recommend than it is to implement.
AI needs to understand how the work fits together
Redesigning a workflow means being clear about the business logic behind it. An organization needs to understand what starts the process, which stages the work passes through, who owns it at each point, what information belongs to it, and where decisions and approvals happen.
It also needs to know who should have access, what should happen next, and how the organization recognizes that the process is complete. These details may feel obvious to the people doing the work, but they’re often based on memory, experience, and informal knowledge rather than anything the organization has defined consistently.
People can usually reconstruct missing context. They know that a document belongs to a particular project even if nothing in the filename says so, and they may remember who normally approves a contract, which version is current, or which email provides the missing piece of information. AI, however, cannot safely depend on that kind of tacit understanding. It needs the relationships between people, information, and processes to be made explicit.
This is why digital groundwork matters. As WorkPoint has explained in its guidance on preparing an organization for Microsoft 365 Copilot, introducing AI also creates an ongoing need for governance, security, and organized information.
Giving AI the context behind the content
WorkPoint creates a shared business layer within Microsoft 365, connecting documents and conversations to the projects, cases, contracts, customers, responsibilities, permissions, and workflows behind them. This gives people and AI a clearer understanding of what information relates to, who owns the work, and where it currently sits within its lifecycle.
Microsoft 365 contains the files, messages, meetings, and other content generated across the organization. WorkPoint gives all of that information a consistent business context. That context allows AI to do more than assist with isolated tasks. It creates the conditions for AI to provide more relevant answers, operate within the correct permissions, support defined workflows, and eventually take action as part of the work itself. Without structure, AI agents guess. With WorkPoint, they know.
WorkPoint is already applying this principle to practical scenarios. We’re already offering ways in which AI can interpret an incoming email, recognize which case or project it belongs to, fill in the project metadata, suggest the next task, and draft a response for review – all within the existing process and permissions. This is a more useful measure of progress than simply counting licences or prompts.
Focus on progress, not the hype
The organizations that gain the most from AI may not be the ones that purchased licences first or generated the most ambitious demonstrations. They are more likely to be the ones that connect the technology to a clear understanding of how their business operates.
That doesn’t mean jumping from occasional Copilot use to a fully agent-operated company. Progress here means asking practical questions about where work currently slows down, what information people struggle to find, and which processes depend on someone remembering what happens next.
Organizations can then consider where AI can remove friction and where human judgement must remain. While these are less dramatic questions than the ones dominating the AI conversation online, they're far more likely to produce useful answers for companies adopting the technology.
To refer to our earlier example, electricity didn’t transform manufacturing simply because factories gained access to it. The larger gains came when organizations redesigned work around its possibilities. AI is following a similar path.
Your organization is probably not as far behind as the noise suggests. Moving forward, however, requires more than another tool; it requires a business environment that both people and AI can understand – the same understanding of your organization that your best people already have locked in their heads.
In the rest of this series, we will explore what that means in practice. We will look at how organizations can help people find the right answers, create the conditions for safe adoption, decide what should be automated, and enable AI agents to move work forward.