Set up an AI agent workflow to automate PM operations

Description

Use this prompt when you want to delegate repetitive PM tasks (ticket triage, backlog grooming, status reporting) to an AI agent that runs autonomously.

You are an AI operations consultant helping me design an agentic workflow to automate repetitive PM tasks.

## Team Context
- Product: {{product_name}}
- Team size: {{team_size}}
- Current tools: {{tools_used}} (e.g., Jira, Linear, Slack, Notion)
- Biggest time sinks: {{time_sinks}}
- Technical comfort level: {{technical_level}}

## Design an AI Agent Workflow

### 1. Task Audit
List the top 10 repetitive PM tasks, ranked by:
| Task | Hours/Week | Automation Potential (High/Med/Low) | Tool Integration Needed |
|---|---|---|---|

### 2. Workflow Design
For the top 5 automatable tasks, design agent workflows:
- **Trigger:** What starts the workflow? (schedule, event, Slack command)
- **Steps:** What does the agent do, in order?
- **Output:** What deliverable does the PM receive?
- **Human checkpoint:** Where does a human review before the agent continues?

### 3. Tool Architecture
- Which AI tools/platforms to use (Claude Code, Zapier, n8n, custom scripts)
- How they connect to existing tools (API, webhook, plugin)
- Data flow diagram: where data enters, transforms, and lands

### 4. ROI Estimate
| Workflow | Hours Saved/Week | Setup Effort | Payback Period |
|---|---|---|---|

### 5. Risk Assessment
- What happens when the agent makes a mistake? (blast radius)
- Which workflows need human-in-the-loop vs. full autonomy?
- Data privacy considerations for AI processing

### 6. Implementation Roadmap
- **Week 1:** Set up one quick-win automation (lowest risk, highest time savings)
- **Week 2-3:** Add 2-3 more workflows with monitoring
- **Month 2:** Evaluate ROI, expand or adjust
- Start with read-only automations before write/send automations

Customize This Prompt