How the boundaries between functions are changing
There is a strange paradox in the modern organization: the better individual functions perform, the more visible the work between them becomes.
Sales can prepare a proposal in minutes, Legal can review a standard contract almost instantly, and Finance can recalculate the economics of a deal on the spot. But the customer outcome does not belong to any one of these functions. It emerges between them.
That space between functions has been one of management’s central challenges for decades. Companies built ERP, CRM, workflow systems and layers of coordination to move work from one functional unit to another.
Now AI is gradually making that handover less inevitable.
This is no longer just a question of individual productivity. It is a question of how work should be organized.
Why organizations are structured this way
The functional organization was a rational response to complexity. Dividing work across Sales, Finance, Legal, Procurement and Operations made it possible to concentrate expertise, establish standards and scale large organizations.
But business outcomes rarely belong to a single function.
Moving a deal from proposal to payment requires several types of expertise. Sales captures the agreement, Legal checks the terms, Finance assesses the economics, and Operations delivers on it. Each owns part of the work — so the work has to pass between them.
Historically, this made sense. Data lived in different systems, specialists worked in different information environments, and the next step required a different type of expertise. This created an infrastructure of coordination: approvals, control points, meetings and procedures.
But if technology can retain the context of a task and execute a sequence of actions across several systems, the underlying assumption starts to change.
Work no longer has to be handed over simply because it crosses a departmental boundary.
When the boundary becomes conditional
Agentic systems can already support processes in which a task moves across several corporate systems without constant manual transfer of context. An agent can receive a request, check the data, access an ERP system, use the corporate knowledge base, prepare a document and trigger the next action.
If a standard deal can move from the commercial decision through financial checks, contracting and execution within a single digital flow, functional departments no longer need to be mandatory stops at every stage.
The question shifts from “how to speed up the handover between functions” to “whether the handover is needed at all.”
This is already emerging in corporate practice. In The Great Enterprise AI Rewrite, Constellation Research describes an Accenture case in which the Contract-to-Cash process was redesigned. The traditional model required around 40 weeks and roughly 50 participants. The new model takes about 10 weeks and involves around 15 people, organized into small teams of five to six.
The important point is not just the speed.
The unit of organizational design is changing.
The process is organized around the outcome — moving from a commercial agreement to cash — rather than around the sequence of departments through which that outcome historically had to pass.
The shift is also visible in wider adoption figures. McKinsey reports that the share of companies using AI across three or more functions has increased from 51% to 56%, while enterprise-wide AI scaling has risen from 38% to 44%.
The Hackett Group and ARIS, in the 2026 Process Context Study, show the other side of the transition: organizations with strong process context are five times more likely to report successful AI outcomes. Yet only 22% of executives report comprehensive real-time visibility into their end-to-end processes, while 59% describe their visibility as fragmented across systems and functions.
Together, these findings point to a growing gap: AI is spreading across functions faster than organizations are changing the way those functions work together.
Organize around work
We see the same pattern when analyzing processes for our clients. Looking at a process end to end, we often find bottlenecks not within functions but between them: where context is handed over, an approval is required, or a task moves into another system.
And this raises a more interesting question: does the work itself still need to follow functional boundaries?
This does not mean eliminating departments. Organizations will still need financial, legal, technology and operational expertise. Some boundaries serve a clear purpose: independent checks, clear responsibilities and risk control.
Functions can act as centers of expertise, joining the flow when their input is needed. Standard scenarios can move through the process, while specialists step in when an exception, higher risk or professional judgment calls for it.
The end-to-end flow of work is gradually becoming the organizational unit, while each function provides expertise and control where they are needed.
For leaders, the practical step is straightforward: instead of launching another AI pilot inside a department, choose one process that crosses several functions and map it backwards from the desired outcome.
Take Contract-to-Cash, for example. At each handover, ask: is it needed for expertise or control, or is it simply part of the way we have always worked?
Keep the checks and expertise the process needs. Redesign handovers that add delay without reducing risk.
And this leads to the next question for the Board and CEO:
If functional boundaries are no longer a technological necessity, which of our organizational rules genuinely protect the business — and which simply preserve a structure that was once necessary?
🔗 Constellation Research. The Great Enterprise AI Rewrite Is Starting. September 2026.
🔗 The Hackett Group & ARIS. 2026 Process Context Study. October 1, 2026.
🔗 McKinsey & Company. The State of AI in 2026: On the Road to ROI. August 25, 2026.
