AI Deployment Is Moving Faster Than Process Redesign

Companies are putting AI into production faster than they can redesign the processes it is expected to improve. In its The AI Process Gap report, Camunda found that 61% said process redesign is moving too slowly to keep pace with AI, while nearly 80% said adding AI to a process creates less internal resistance than redesigning it.

That pattern carries a cost: 72% said process-related challenges contributed to failed AI initiatives, at an average reported cost of $1.55 million. The findings point to process design as one factor that can determine whether AI reduces work or shifts it elsewhere.

The report is based on surveys of 1,000 process decision-makers and 5,000 employees at organizations with at least 1,000 employees in the US, UK, France, and Germany.

Why Existing Processes Can Limit What AI Changes

Camunda uses “legacy process” to describe more than outdated technology. A process can be current and still reflect assumptions that predate AI, with work organized around human review, sequential approvals, and handoffs between teams.

Adding AI does not automatically change that structure. An agent may complete one step more quickly while the surrounding approvals and handoffs remain in place. The time saved at one point can then be lost elsewhere in the workflow.

Task-level speed can hide delays elsewhere in the process. The more useful measure is how the end-to-end workflow performs after AI is introduced.

Where AI Creates More Work

Leaders and employees see the productivity impact differently. While 89% of leaders said AI is making their teams more productive, only 64% of employees agreed.

Forty-four percent of employees said they had manually overridden AI outputs because the underlying process was not set up correctly. Another 33% said AI had increased their workload. In some cases, the effect was significant enough to reverse course: 48% of organizations said they had rolled back AI use after it negatively affected employees’ ability to do their jobs.

Process weaknesses also show up in governance. Forty percent of organizations reported an AI-related compliance or governance issue during the previous 12 months. Among that group, 84% said process-related challenges contributed to the incident. Camunda points to misplaced controls, undocumented handoffs, and incomplete audit trails as examples of where exposure can develop.

What Process Re-Engineering Changes

The research also points to a mismatch in where AI budgets are going. Thirty-seven percent of organizations said they devote 10% or less of their AI spending to process redesign and the organizational change around it. At the same time, 82% said their AI investments will fall short without more investment in process design.

Camunda calls its response “The Great Process Re-Engineering.” The approach starts with the result a process is supposed to produce, then reconsiders how the work should move once AI is available. That can change where human review occurs, which steps remain necessary, and how responsibility moves between people and software.

ProcessOS applies AI to the redesign work itself. Camunda has developed four specialized agents that cover discovery, re-engineering, build and deployment, and continuous improvementv. The product can reconstruct how a process currently runs from operational data and documentation, then generate a proposed future-state process along with supporting integrations, decision logic, and agent prompts.

Camunda says ProcessOS can test a proposed redesign against historical process performance before it moves into production, with human approval remaining the final gate. AI can continue proposing changes after a process is deployed, but enterprise teams still need evidence that those changes improve how the workflow performs. ProcessOS is still in closed beta, but it shows how Camunda expects that review cycle to work.

What This Means for SAPinsiders

• AI makes bad processes more expensive. When agents take on more work, existing delays and weak handoffs can spread across the workflow faster. Problems that once affected one team can become a broader operating cost.
• AI budgets need room for process change. Companies can fund the technology and still underfund the work needed to make it useful. Process redesign should be treated as part of the investment, not an afterthought.
• Process owners will play a bigger role in AI governance. As agents take on more decisions and actions, someone still has to define where automation stops and human judgment begins. That puts process ownership closer to the center of AI control.