Using Deep Work
Deep Work offers you a few different ways to run your prompts:
Suggested Prompts - these appear below the input bar and are updated frequently to give you customized and relevant suggestions
Agent Templates - these are tailored frameworks that simplify your input requirements and leverage pre-defined paths to generate your output more effectively
Custom Prompts - you have the freedom to run any prompt you would like. See below for some helpful tips for optimizing your output
Step 1: Define Your Objective
Be clear about what you want to accomplish:
What decision are you trying to make?
What information do you need?
Who will use the output?
Step 2: Specify Your Scope
Include relevant parameters:
Timeframe: "over the last quarter", "from last week", "in the past month"
Data sources: "from Slack and meetings", "across all customer interactions"
People/teams: "for the engineering team", "regarding customer SafetyWing"
Step 3: Request Specific Deliverables
Be explicit about the format you need:
"Provide a summary with key takeaways"
"Create a report I can share with executives"
"Generate action items and next steps"
"Include metrics and supporting data"
Best Practices for Deep Work Prompts
✅ Do This
Be specific: "Analyze Q3 sales performance" vs "Tell me about sales"
Include context: "For our board meeting next week"
Set boundaries: "Focus on the top 3 issues"
Specify format: "Create a structured report with recommendations"
❌ Avoid This
Vague requests: "Tell me what's happening"
No timeframe: "Look at our customers"
Too broad: "Analyze everything"
No clear purpose: "Just curious about..."
Quick Start Examples
Try these prompts to get familiar with Deep Work:
Meeting Summary: "Summarize my meetings from yesterday and highlight any action items assigned to me"
Team Update: "What's happening around my org that I might need to know?"
Customer Insight: "What are the top 3-5 use cases for [customer name] that I should add to our tracker?"
Project Status: "What are my top priority tasks today?"
Troubleshooting
If Your Results Aren't What You Expected
Too broad? Add more specific parameters and constraints
Missing information? Specify additional data sources or timeframes
Wrong format? Be more explicit about the deliverable you need
Not actionable? Ask for specific recommendations and next steps