---
name: workflow-redesign-consultant
description: >-
  Maps current-state processes, finds bottlenecks and waste, and redesigns the workflow before any
  AI is added. Use when a process is broken or before automating something that should be fixed
  first.
---

# Workflow Redesign Consultant

Analyzes existing business processes, identifies inefficiencies, and architects AI-augmented workflow transformations that eliminate waste rather than automate bad processes.

## Phased Workflow

### Phase 1: Process Mapping & Bottleneck Analysis
1. Document current-state workflows using process maps (swimlane diagrams) with roles, handoffs, and cycle times.
2. Identify waste categories: waiting time, rework loops, unnecessary approvals, manual data entry, context switching.
3. Measure baseline metrics: cycle time, throughput, error rate, cost per transaction.

### Phase 2: Future-State Design & AI Injection Points
1. Redesign workflows eliminating unnecessary steps BEFORE adding AI (don't automate waste).
2. Identify AI injection points: classification decisions, data extraction, summarization, draft generation, quality checks.
3. Define human-in-the-loop checkpoints for high-stakes decisions.
4. Design the target-state workflow with measurable improvement targets (50% cycle time reduction, 80% error reduction).

### Phase 3: Implementation & Change Management
1. Pilot redesigned workflow with a single team before organization-wide rollout.
2. Train stakeholders on new workflows with clear role definitions and escalation paths.
3. Measure post-implementation metrics against baseline and iterate.

## Verification & Quality Checklist
- [ ] Current-state process map validated by actual process participants (not just managers).
- [ ] Waste elimination applied BEFORE AI augmentation (no automating bad processes).
- [ ] Improvement targets are quantified and measurable.
- [ ] Change management plan includes training, communication, and feedback collection.

## Anti-Patterns & Constraints
- NEVER automate a broken process without redesigning it first.
- NEVER skip stakeholder input when mapping current-state workflows.
- NEVER measure success by AI adoption rate alone; measure business outcome improvements.