Using AI Suggestions
This tutorial shows you how to use HeroJournal’s AI features effectively to write better achievement descriptions with less effort.
Prerequisites
Before starting, ensure:
- AI is enabled in Settings > AI
- You have an API key configured (or Ollama running locally)
- Connection test passes
See AI Integration for setup instructions.
Workflow 1: Generate from Title
The quickest way to use AI - let it generate all altitude levels from just a title.
Step 1: Create Achievement with Good Title
- Click New Achievement
- Enter a descriptive title:
Good titles:
- “Launched checkout redesign increasing mobile conversion 25%”
- “Built real-time analytics pipeline processing 1M events/day”
- “Led security audit achieving SOC 2 compliance”
Weak titles (avoid):
- “Checkout project”
- “Analytics work”
- “Security stuff”
Step 2: Add Context (Optional but Helpful)
In the Notes field, add context the AI can use:
- Background on the project
- Team size and your role
- Any specific metrics you know
- Challenges faced
Step 3: Generate Each Level
For each altitude level:
- Click the Generate button (sparkle icon)
- Wait for AI to generate text
- Review and edit as needed
- Move to next level
Step 4: Refine
After generating all levels:
- Add specific details AI couldn’t know
- Correct any factual errors
- Adjust tone to match your voice
Workflow 2: Improve Existing Text
When you have a rough draft and want to polish it.
Step 1: Write Your Initial Version
Don’t worry about perfection. Write what you know:
“We built a new checkout flow. It was faster and people liked it. Sales went up.”
Step 2: Use Improve
- Click the Improve button (wand icon)
- AI refines the text while preserving your meaning:
“Redesigned the checkout flow, reducing steps from 5 to 3. The streamlined experience improved user satisfaction scores by 12% and increased conversion rate by 25%.”
Step 3: Keep What Works
- Compare original and improved versions
- Keep improvements you like
- Revert parts that don’t sound like you
- Add missing specifics
Workflow 3: AI Chat for Feedback
Use the chat panel for interactive help.
Opening the Chat
Click the chat icon in the right sidebar to expand the AI panel.
Asking for Feedback
With an achievement open, try:
- “Is this too technical for my manager?”
- “How can I quantify this better?”
- “What metrics might I be missing?”
- “Is the business impact clear?”
Asking for Alternatives
- “Give me 3 different ways to phrase this”
- “How would I explain this to an executive?”
- “What’s a stronger opening for this description?”
Getting Suggestions
- “What evidence should I include?”
- “What metrics typically apply to this type of work?”
- “How can I show leadership in this achievement?”
Best Practices
Do’s
- Provide context - The more you give, the better the output
- Include numbers - AI works better when you mention metrics
- Be specific in requests - “Make this more technical” vs “Improve this”
- Iterate - Generate multiple versions and combine the best parts
- Verify facts - AI can make things up; check all specifics
Don’ts
- Don’t accept blindly - Always review and edit AI output
- Don’t over-rely on AI - Use it as a starting point
- Don’t share sensitive info - Be mindful of what context you provide
- Don’t expect perfection - AI helps, but your knowledge is essential
Example: Full AI-Assisted Achievement
Let’s walk through creating an achievement with AI help.
Start: Title and Notes
Title: “Built customer feedback system reducing support tickets”
Notes: “Created in-app feedback widget. Integrated with Zendesk. Took 6 weeks. Team of 3 engineers. Support tickets about confused customers dropped after.”
Generate: For Peers
AI generates:
Designed and implemented an in-app feedback widget using React and a custom API, integrated with Zendesk for ticket management. The system includes real-time sentiment analysis and automatic categorization of feedback.
You edit (adding specifics):
Designed and implemented an in-app feedback widget using React with a Node.js backend API. Integrated with Zendesk using their REST API for automatic ticket creation. Implemented real-time feedback aggregation in Redis for the analytics dashboard.
Generate: For Manager
AI generates:
Led a 3-person team over 6 weeks to deliver the customer feedback system, resulting in a measurable reduction in support tickets. The streamlined feedback process improved customer satisfaction while reducing support team workload.
You edit (adding metrics):
Led a 3-person team over 6 weeks to deliver the customer feedback system. The in-app widget increased feedback submission by 300% while reducing confused-customer support tickets by 35%, saving an estimated 20 hours/week for the support team.
Generate: For Skip Level
AI generates:
The feedback system directly supports our customer experience OKR by providing a scalable channel for customer input. Cross-team collaboration with Support and Product ensures feedback informs roadmap decisions.
You edit:
Directly supports Q4 Customer Experience OKR by creating a direct voice-of-customer channel. The system is now used by Product to inform roadmap decisions and by Support to proactively identify product issues before they escalate.
Generate: For Leaders
AI generates:
Feedback system reducing support costs and improving customer experience metrics.
You edit (adding specifics):
Built customer feedback channel that reduced support costs by $50K annually while increasing NPS score by 8 points.
Troubleshooting
Poor Quality Suggestions
- Add more context in Notes
- Use a more descriptive title
- Try a different AI model
- Be more specific in your request
AI Doesn’t Understand Context
- Fill in more achievement fields first
- Use the chat to explain context
- Generate multiple times for variety
Suggestions Are Too Generic
- Include specific technologies/tools
- Mention team size, timeline, scope
- Add known metrics even as rough estimates
Time-Saving Tips
- Batch processing - Generate all levels for multiple achievements in one session
- Templates - Note phrasing patterns you like for future use
- Quick notes - Keep rough notes when work happens; AI can polish later
- Chat shortcuts - Save common prompts for reuse