Prompt Improver
The Prompt Improver analyzes your input and enhances it with better structure, context, and specificity—automatically detecting when you’re working with construction and engineering content.
The Prompt Improver works best with prompts that have a clear intent but lack structure. Quick factual questions (“What’s the IBC fire rating for corridors?”) usually don’t need improvement.
What It Does
Quality Analysis
Detects if your prompt is sparse, moderate, or complete—and adjusts enhancement accordinglyAEC Domain Detection
Automatically recognizes construction terms like RFI, submittal, BIM, and CSI divisionsStructured Output
Adds context, constraints, and expected output format to get more focused responsesBefore and After Examples
Example 1: Sparse to Structured
| Before | After |
|---|---|
| ”find hvac specs” | Context: Project documentation search for mechanical specifications. Task: Locate HVAC equipment specifications including make, model, capacity, and efficiency ratings. Constraints: Focus on rooftop units and air handling units. Expected Output: Table with equipment tag, manufacturer, model number, and key specifications. |
Example 2: AEC Domain Enhancement
| Before | After |
|---|---|
| ”write an RFI response about the concrete” | Context: Construction project RFI response for concrete work. Domain: AEC (Architecture, Engineering, Construction). Task: Draft a professional RFI response addressing concrete-related questions. Constraints: Follow standard RFI format with clear, actionable information. Reference applicable specifications and codes. Expected Output: Formal RFI response document ready for submittal. |
When to Use Prompt Improver
Good Candidates
- Research-heavy queries that need multiple data points
- Document generation requests (specs, reports, meeting minutes)
- Complex multi-step analysis tasks
- Queries spanning multiple trades or systems
Skip It For
- Simple factual questions
- Follow-up questions in an existing conversation
- When you’ve already structured your prompt carefully
How Enhancement Works
The Prompt Improver analyzes your prompt across several dimensions:
- Intent Detection - What are you trying to accomplish?
- Domain Recognition - Is this construction/AEC related?
- Completeness Check - What context is missing?
- Output Specification - What format should the response take?
Based on this analysis, it generates an enhanced prompt that:
- Provides necessary context
- Specifies constraints and scope
- Defines the expected output format
- Uses appropriate industry terminology
Tips for Better Improvements
Provide Intent
Even a few words help the improver understand your goal:
❌ “specs” → Too vague ✅ “find specs” → Shows you’re searching ✅ “write specs” → Shows you’re generating
Include Key Terms
Industry-specific terms help with domain detection:
❌ “equipment list” ✅ “HVAC equipment schedule” ✅ “Division 23 mechanical equipment”
Mention Project Context
If relevant, include project or phase information:
❌ “RFI response” ✅ “RFI response for concrete curing, Civic Center project”
Save Improved Prompts to Your Library
Once you’ve improved a prompt that works well, save it to your Prompt Library for reuse. This turns one-time improvements into permanent time-savers.
How to Save an Improved Prompt
Example: Saving an RFI Prompt
Starting prompt: “write rfi for missing door hardware”
After improvement:
Context: Construction project RFI for architectural hardware.
Domain: AEC (Architecture, Engineering, Construction).
Task: Draft a professional RFI requesting clarification on missing door hardware specifications.
Constraints: Reference the affected door schedule locations. Include impact to installation timeline. Follow standard RFI format.
Expected Output: Formal RFI document with specific questions about hardware sets, finish, and manufacturer.Save it as:
- Title: “RFI Draft - Missing Specifications”
- Category: RFIs, Generation
- Description: “Creates professional RFIs for missing or unclear specification items”
Now you can apply this prompt anytime you need to draft an RFI for missing specs—no improvement step needed.
Next Steps
Was this page helpful?