Guide: Process Automation with the AI Assistant
Introduction
With this guide, you will develop your own automation roadmap – step by step, supported by the AI Assistant.
The assistant helps you identify processes that still contain manual activities and highlights specific automation opportunities.
Simply click on a prompt to use it directly in the AI Assistant chat – from an initial process overview to concrete automation ideas.
Step 1 – Create an overview
📋 Example Output:
| Process | Main Tasks |
|---|---|
| Car Rental | Manage the process from booking to return |
| Vehicle Maintenance | Inspect damage, clean, plan maintenance |
| Customer Complaint Management | Record, process, and resolve complaints |
| Customer Support | Handle, escalate, and document customer requests |
👉 Goal: Build a clear process landscape – even without prior knowledge.
AI Assistant Logic
- Analyzes existing process information or uses your input in chat.
- Detects common business areas and groups related steps into main processes.
- Summarizes tasks in a clear, easy-to-understand process overview.
Step 2 – Focus on manual activities
📋 Example Output:
| Process | Activity | Status |
|---|---|---|
| Customer Complaint Management | Review incoming emails | manual |
| Customer Complaint Management | Draft response | manual |
| Customer Support | First contact & triage | manual |
| Vehicle Maintenance | Execution & quality check | manual |
| Vehicle Maintenance | Cleaning | semi-automated |
👉 Goal: Identify manual or partially automated process steps.
AI Assistant Logic
- Breaks down main processes into atomic steps (e.g., review, record, send).
- Assesses each step based on its automation level (manual / semi / automated).
- Recognizes recurring patterns from similar processes to estimate automation status realistically.
Step 3 – Identify frequency & impact
📊 Example Output:
| Sub-step | Frequency | Effort per case | Impact (F×E) |
|---|---|---|---|
| Draft response | high | high (10–20 min) | ⭐⭐⭐ very high |
| Read & record complaint | high | medium–high (5–10) | ⭐⭐⭐ very high |
| Forward to department | medium | medium (3–5 min) | ⭐⭐ high |
| Send response | high | low–medium (2–4) | ⭐ medium |
| Review incoming mail | high | low (1–2 min) | ⭐ medium |
👉 Goal: Understand where the highest automation potential lies.
AI Assistant Logic
- Combines process frequency with effort per case.
- Calculates an impact score (frequency × effort).
- Sorts results by leverage – highest priority first.
Step 4 – Identify quick wins
⚡ Example Output:
Top 3 short-term automations (2–4 weeks):
-
Review incoming mail → Auto-ticket & auto-acknowledge
🛠 Solution: Mail gateway or Graph/Gmail API → automatically create ticket + send acknowledgment.
⏱ Effort: ~1 week.
✅ Effect: Faster first response, fewer missed emails. -
Forward to department → Rule-based routing
🛠 Solution: If-Then rules in CRM/ticketing system → automatic assignment + notification.
⏱ Effort: ~1 week.
✅ Effect: Fewer misroutings, shorter cycle times. -
Send response → Templates + approval workflow
🛠 Solution: Standard templates with placeholders + 4-eye review.
⏱ Effort: 1–2 weeks.
✅ Effect: 15–25% less handling time, more consistent quality.
👉 Goal: Make tangible improvements visible immediately.
AI Assistant Logic
- Filters the highest-impact steps from Step 3.
- Evaluates technical simplicity (rules, APIs, RPA) and delivery time.
- Creates a shortlist of feasible quick wins (2–4 weeks).
Step 5 – Develop automation ideas
🤖 Example Output:
Top automation opportunities in customer complaint management:
- Review incoming mail – Auto-ticket & auto-acknowledge (Mail API + rules); AI classification: complaint vs. inquiry
- Read & record complaint – Auto-field population (OCR/Document AI); extract entities & risk flags
- Forward to department – Rule-based routing (queues/teams); ML-based routing for unclear cases
- Draft response – Templates + knowledge base; LLM-assisted response generation with tone check
- Send response – Auto-send with audit trail; AI quality check before dispatch
👉 Goal: Provide concrete, actionable solutions – not just theory.
AI Assistant Logic
- Matches prioritized steps with suitable technologies (OCR, APIs, LLM, rules).
- Uses best practices and patterns to propose combined solutions.
- Assesses impact, effort, and technical feasibility for each proposal.
Step 6 – Prioritize
📊 Example Output:
| Process Area | Benefit | Feasibility | Quick Wins (Examples) |
|---|---|---|---|
| Customer Complaint Management | ⭐⭐⭐ very high | ⭐⭐⭐ high | Auto-ticketing, routing, templates, LLM drafts |
| Customer Support | ⭐⭐ medium–high | ⭐⭐⭐ high | AI triage, macro responses, KB retrieval |
| Car Rental | ⭐⭐ medium | ⭐⭐ medium–high | ID/license check, fee rules |
| Vehicle Maintenance | ⭐ medium | ⭐ medium | Service workflows, recall triggers |
👉 Goal: Provide a clear decision basis for management.
AI Assistant Logic
- Assesses benefit (efficiency, cost, quality) and feasibility (technical, organizational).
- Weights both factors into an overall score.
- Recommends processes with the highest leverage and lowest complexity first.
Step 7 – Summarize business impact
🎯 Example Output:
- Efficiency & Speed – Faster response times (auto-ticketing, routing, templates/LLM); fewer handoffs & follow-ups; better SLA compliance
- Cost Reduction – Less handling time per case; scalable without extra staff; fewer reworks
- Quality Improvement – Consistent outcomes, higher data quality; validations & compliance checks; audit trails
- Employee Relief – Eliminate repetitive work; focus on complex tasks & customer care; fewer context switches
- Customer Experience – Faster initial responses; higher solution quality; improved satisfaction (CSAT/NPS)
- Control & Transparency – Live KPIs on volume, timing, bottlenecks; better forecasting & planning decisions
👉 Goal: Build a clear, data-driven business case.
AI Assistant Logic
- Aggregates effects from previous steps into main benefit categories (efficiency, cost, quality, employee, customer, control).
- Quantifies improvements when data is available (e.g., minutes per case, € per transaction, NPS points).
- Creates a clear, management-ready value summary.
Step 8 – Generate PDF Business Case Report
👉 Goal: A concise, management-ready PDF report summarizing all benefit dimensions (efficiency, cost, quality, employee, customer) clearly, quantitatively, and visually.
AI Assistant Logic
- Uses the results from the benefit analysis to automatically generate a structured management report.
- Summarizes all improvements (efficiency, cost, quality, employee, customer) in clear tables with key metrics.
- Includes a short justification for each metric and highlights improvements visually (📈 + percentage values).
- Formats the report using corporate design (A4 layout, colors, fonts) and creates a print-ready PDF file.
- Closes with a concise executive summary and conclusion for management decision-making.
Conclusion
With these prompts, the AI Assistant guides you from an initial overview to a full business case for process automation:
- Steps 1–3: Identify manual activities and quantify their effort
- Steps 4–5: Develop quick wins and concrete automation solutions
- Steps 6–7: Derive benefits, priorities, and a management-ready business case
🎯 Result: A clear, prioritized automation roadmap – including quick wins, impact estimation, and a solid business case to support decision-making.
👉 Get started now: Open the prompts in chat, apply them to your own processes – and gradually achieve greater efficiency, quality, and transparency through targeted automation.