| name | fde-reengineer-process |
| description | Map an evidence-backed current-state business process and redesign it across simplification, deterministic automation, UiPath RPA, agentic AI, human-in-the-loop, and human-only work. Use for process mapping, exception analysis, AI suitability, automation allocation, future-state design, adoption analysis, or when a proposed automation risks being layered onto a broken process. |
Reengineer an FDE Process
Design from evidence and exceptions, not the documented golden path.
Gate the work
- Read the engagement charter, status, knowledge needs, contradictions, evidence index, stakeholders, and risks.
- Read
../fde-run-engagement/references/lifecycle-gates.md.
- Invoke
$fde-interview-engagement for gate-critical gaps.
- Invoke
$fde-capture-knowledge when source reconciliation is incomplete.
- Stop future-state design until the current-state boundary and high-risk contradictions are owned.
Completion criterion: The process boundary, owner, objective, trigger, end condition, and evidence status are explicit.
Model the current state
For every step record:
ID | name | performer | accountable owner | trigger | inputs | systems |
action/decision rule | outputs | downstream dependency | normal duration |
volume | exceptions | escalation | controls | evidence | confidence/status
Trace normal flow and each material exception. Identify rework loops, handoffs, queues, wait states, manual workarounds, control points, duplicate data entry, and undocumented owners.
Use $uipath-process-domain-modeling when terms or business concepts are unstable. Use the relevant official UiPath artifact skill only after the process knowledge is ready.
Challenge the process
For each step ask in order:
- Can the outcome or step be eliminated?
- Can policy or data be simplified first?
- Is the behavior deterministic enough for an API/workflow/RPA implementation?
- Does it require probabilistic interpretation or judgment suitable for an agent?
- Which decisions need a human checkpoint?
- Which step remains human-only because of accountability, empathy, negotiation, physical work, or unacceptable failure impact?
Record the disposition in fde/process/automation-allocation.md as:
- eliminate/simplify;
- deterministic automation;
- UiPath RPA/API workflow;
- agentic AI;
- human-in-the-loop;
- human-only.
Justify using volume, variability, data quality, judgment, reversibility, explainability, control obligations, failure severity, adoption cost, and measurable value.
Design the future state
Update fde/process/future-state.md with:
- new sequence and dependencies;
- retained human responsibilities;
- automation/agent responsibilities;
- approval and escalation points;
- source-of-truth systems;
- error recovery and fallback;
- observability and outcome measures;
- adoption/training impact;
- migration assumptions.
Compare current and future states step by step. No current step disappears without a recorded disposition.
Completion criterion: Every step, exception, control, and ownership change is accounted for, measurable, and traceable to evidence or an approved decision.
Finish
Return the current-state risks, automation allocation, future-state value hypothesis, unresolved design decisions, required approvals, and the next official UiPath design skill.
