The Capability
Zurich release includes Process and Task Mining — AI reviews actual workflow execution logs, identifies patterns (delays, loops, bottlenecks), recommends improvements. Closes the gap between process design and process reality.
What It Finds
Tasks that take 10x longer than expected. Loops where workflow bounces between users. Approvals that rubber-stamp 95% of time (candidate for auto-approval). Paths that never get taken (dead branches).
Acting on Recommendations
Review mining output monthly. Prioritize recommendations by frequency × impact. Implement via workflow tweaks or agentic playbook additions. Track impact post-change — did the inefficiency actually reduce?
Limits
Correlation not causation. Mining shows what happens; humans interpret why. Some ‘inefficiencies’ are required checks. Review carefully before automating away what looks redundant but matters for compliance or quality.