Turn AI-agent work into a UiPath-governed case with evidence, risk, approval, robot work items, and human handoff.
When an agent touches production systems, teams need to know what happened, what evidence was used, what risk was found, who approved it, and how work can be resumed safely.
What did the agent do?
Was a risky action blocked?
Who approved the final decision?
Case stages: intake, agent investigation, robot/API evidence, risk review, approval, handoff.
Evidence collection is modeled as queue-ready robot work items.
Human approval tasks preserve evidence, risk, and decision context.
Implementation artifacts include a BPMN-style process blueprint, case data model, Action Center form schema, robot queue model, state machine, transaction logs, and simulated case run.
uipath-package/maestro-process.bpmn
runtime/case-state-machine.json
runtime/orchestrator-transaction-log.jsonl
cd uipath-agenthack
bash scripts/run_uipath_local_checks.sh
verify_ok
uipath_verify_ok
AgentOps Case Control Room shows how human, AI, robot, and API work can be reviewed as one case. That is the bridge between a prototype on a laptop and agentic automation that enterprises can trust.