Putting capability where knowledge lives
When an insurance underwriter evaluates a claim, she draws on years of pattern recognition. Edge cases she’s seen, exceptions she’s learned, judgment calls that never made it into any manual. When her company decides to automate that process, she hands off requirements to IT and waits. What arrives weeks later handles the straightforward cases. The rest still needs her.
And despite billions invested in automation platforms, this pattern persists across companies. Requirements get simplified in translation. Edge cases get omitted because they seemed obvious to the person who didn’t write them down. The judgment that lives in experienced heads stays there.
The solution isn’t more time spent on requirements gathering. It’s rethinking who builds automation in the first place.
I believe we’re at a turning point. The center of gravity for automation has been shifting from IT toward the business, and organizations that have recognized and are adapting to this shift will pull ahead of those that don’t.
The automation ownership problem
For decades, automation has been IT’s domain. And for good reason. Building workflows meant writing code, understanding system architectures, and managing complex dependencies. IT owned it because the work demanded technical expertise most business teams didn’t have.
But that model has a fundamental weakness. Those who understand the business aren’t those building the automation.
What’s obvious to the business user never makes it into the requirements. Edge cases that an experienced professional handles instinctively become impossibly complex to document. The logic built over years of pattern recognition gets reduced to a simplified specification that misses the nuances.
The result is automation that handles the predictable 80 percent and breaks on the 20 percent that matters.
I’ve spoken with numerous enterprise leaders over the past year. The CIO of a major financial services firm put it simply: “There are ten times more business users than IT users. Why are we asking IT to automate what the business knows best?”
Why low-code and no-code fell short
The industry recognized this problem years ago. The response was low-code and no-code platforms. Reduce the technical barrier, make automation accessible to more people.
It was the right instinct. But the execution fell short.
Low-code solutions reduced some complexity while still requiring technical thinking. Business users hit walls when processes got sophisticated. No-code alternatives achieved simplicity by sacrificing capability. Fine for basic tasks, but inadequate for the complex workflows that drive real business value.
Neither approach solved the deeper problem: Brittleness. Even when business users could build workflows, those workflows still broke the moment conditions changed.
Traditional automation operates on deterministic logic. It works perfectly when data arrives in the expected format, when processes follow predicted paths, when nothing changes. The moment reality deviates—an unexpected field, a new scenario, a shifted business rule—it breaks.
A 2025 EMA research study of over 400 companies found that the complexity of managing automation across hybrid and multi‑cloud environments is one of the top reasons organizations struggle to expand automation. Companies don’t trust their automation to handle variability in complex environments. So, they limit scope, maintain manual fallbacks, and never achieve the transformation they projected.
The gap between what business teams need and what they can build hasn’t closed.
How AI changes business automation
Generative AI is closing that gap. Not by replacing human judgment, but by removing the barriers that have kept business teams from participating.
The shift happens across three dimensions:
- Workflow creation – Natural language interfaces allow practitioners to describe what they need conversationally, with AI translating intent into workflow logic. Someone who understands the business process but lacks technical training can build automation that previously required specialized skills.
- Operational visibility – AI-powered assistants can answer questions like “which workloads failed in the past 24 hours?” by analyzing logs, metrics, and run history in real time. Knowledge that once required navigating multiple systems becomes available through conversation.
- Adaptability – AI accommodates variability in ways deterministic systems cannot. Data arriving in unexpected formats, edge cases that weren’t explicitly programmed. Scenarios that would break traditional automation become manageable variations.
None of these capabilities replaces human judgment. Rather, those who understand the business can finally work directly with systems that were once the exclusive domain of IT.
How organizations are adopting business-driven automation
Organizations are changing how they evaluate and select automation. Traditional technology-silo approaches—picking the best tool for each category—are giving way to business-outcome evaluations where multiple automation capabilities are assessed together against the problems they need to solve.
Ease of use for business teams is becoming a standard selection criterion. Companies want tools their operations teams can configure without filing IT tickets. They want dashboards that business users can customize instead of waiting for reports. They want workflows that business analysts can safely modify when requirements change.
Some enterprises have gone further, embedding IT specialists within business functions to create hybrid teams that blend technical capability with domain expertise.
But the most promising path is technology that reduces the expertise required in the first place. AI-powered workflow creation that lets practitioners describe what they need in natural language. Conversational interfaces that bring expertise directly into daily work—whether that’s mainframe knowledge for application developers or operational insights for business teams. Intelligence embedded where work already happens, not in separate tools that require separate training.
The future of business-driven automation
The path forward isn’t about choosing between IT control and business freedom. It’s about creating systems sophisticated enough to handle real complexity while accessible enough for business teams to own.
Three principles guide this shift:
- Put capability where knowledge lives – The people who understand outcomes should have the tools to act on that understanding directly.
- Embed intelligence in existing workflows – AI that requires separate tools and training adds friction. AI that meets practitioners where they work removes it.
- Design for variability, not just efficiency – Automation that breaks when reality deviates isn’t automation. It’s a liability.
These aren’t aspirational goals. They’re practical requirements for organizations that want automation to deliver on its promise.
The underwriter who knows exactly how to assess risk should be building the automation that does it. The operations manager who understands workflow dependencies should be configuring how they run. The finance team that sees patterns in the data should be creating the rules for how to act on them.
The question is whether organizations will embrace the shift or keep waiting for IT to translate what the business already knows.
