Salesforce recently highlighted a powerful idea: make existing enterprise capabilities and business logic easier for AI agents to reuse.
That is useful technology.
But before organizations make existing logic easier for agents to consume, there is a more basic question:
A workflow being established does not mean it is efficient. A policy being encoded does not mean it still makes sense. An approval existing for years does not mean it still protects anything of value.
Organizations accumulate process debt just as they accumulate technical debt.
Extra approvals get added after one bad incident. Handoffs develop around organizational boundaries. Workarounds become standard operating procedure. Policies remain long after the conditions that created them disappear.
Now imagine making all of that easier for AI agents to execute at scale.
AI accelerates the system you give it. If the underlying operating model is sound, that can create enormous value. If it contains unnecessary complexity, redundant controls, poor decision rights, or inefficient workflows, AI can accelerate those problems too.
Deloitte's recent research shows how significant this gap may be. Seventy-four percent of surveyed leaders expect nearly half of their business processes to be redesigned or rebuilt around AI agents within four years. Yet only 5% say their business processes are highly prepared for agents today. Deloitte also found that 75% believe human collaboration with AI agents creates more value than automation alone.
Those numbers point to something bigger than an AI-readiness problem.
They point to a process-readiness problem.
Before organizations expose more business logic to agents, they should examine what that logic actually represents.
- Why does this approval exist?
- Is this handoff necessary?
- Does this policy still support the intended outcome?
- Is the decision being made at the right point in the process?
- Is the workflow designed around a technology limitation that no longer exists?
And perhaps most importantly:
Salesforce is making reuse easier. Deloitte is highlighting how unprepared many organizations remain at the process level.
Taken together, they make the same underlying point clear:
Greater technical capability does not eliminate the need for process judgment. It makes that judgment more important.
The agentic era is beginning to turn business logic into reusable infrastructure. Before we scale that infrastructure, we should make sure we are not simply preserving yesterday's operating model and teaching tomorrow's technology to run it faster.
We spent decades asking whether technology could execute our processes. AI agents are giving us a reason to ask the better question:
Are these the processes we want technology to execute?