OpenHuman Prompt Tips — How to Ask Your AI Assistant the Right Way

The quality of OpenHuman's answers depends largely on how you ask. These prompt engineering tips will help you get more accurate, relevant, and useful responses from your AI assistant.

1. Be Specific About Context

OpenHuman has access to your Memory Tree, emails, calendar, and code. Tell it where to look:

• ❌ "What meetings do I have?"
• ✅ "What meetings do I have this week from my Google Calendar?"
• ❌ "Summarize my emails"
• ✅ "Summarize the important emails from today, focusing on anything from my team"

2. Leverage the Memory Tree

OpenHuman remembers past conversations. Reference them:

• "Based on our earlier discussion about the marketing plan, what were the key action items?"
• "Remember the project I mentioned last week about the dashboard redesign?"

3. Specify Output Format

Tell OpenHuman how you want the answer:

• "List the key points as a bullet list"
• "Give me a table comparing the three options"
• "Summarize in 3 sentences"
• "Write this as a draft email"

4. Use Follow-Up Questions

OpenHuman maintains conversation context. Build on previous answers:

• You: "What PRs are open in my repo?"
• AI: Lists PRs...
• You: "Summarize the changes in the second one"

5. Combine Integrations

Take advantage of OpenHuman's multi-service awareness:

• "Check my calendar for free slots this week, then find the email thread about the project review, and draft a meeting invite"
• "Look at the Slack discussion from yesterday about the bug fix and check if there's a related GitHub issue"

6. Debug Mode

When troubleshooting, ask OpenHuman to reason step by step:

• "I'm getting an error when I try to connect Slack. Walk me through the troubleshooting steps."
• "Why isn't my Ollama model responding? Check my config and suggest fixes."

7. Don't Over-Explain

OpenHuman already has context from your Memory Tree and integrations. Let it surprise you:

• "What should I focus on today?" — OpenHuman uses your calendar, emails, and PRs to suggest priorities
• "Is there anything urgent I missed?" — cross-references all integrations