Your First AI Application
This walkthrough will take you from idea to deployed AI application in just a few minutes using Empromptu's Builder.
What you'll build ⏱️ 5 minutes
A Review Summarizer that analyzes customer feedback and extracts key insights. You'll learn how to:
Use the Builder to create your application
Understand the project and task structure
Run your first optimization
Deploy your application
Step 1: Access the Builder
Navigate to the Builder section in your Empromptu dashboard. You'll see the main interface with the heading "Let’s build something together."
The Builder interface includes:
A text input asking "What are you building?"
Example prompts to get you started
Recent chat history (if you've used the Builder before)
Step 2: Describe Your Application
In the "What are you building?" field, type:
Build a Review Summarizer that analyzes customer reviews and extracts key insights, emotional reactions, and overall sentiment.Pro tip: Be specific about what you want your application to do. The more detail you provide, the better Empromptu can build your application.
Press Enter or click the send button. Empromptu's AI will start building your application automatically.
Step 3: Continue Building or Access LLMOps
After the Builder completes (usually within 30-60 seconds up to a few mins), you have two options:
Continue Building in Builder
Stay in the Builder interface to refine your application, add features, or make adjustments.
Improve your application using our LLMOps tooling
Click the Optimize tab on the left sidebar to access your project's optimization tools and analytics.
Step 4: View Your Project Dashboard
Click the Projects tab on the left sidebar to see your project dashboard:
Project Overview
Review Summarizer A project that analyzes customer reviews and extracts key insights.0.00% 0.00% 1Average Initial Accuracy Average Current Accuracy Total TasksProgress: 0.00%Tasks Table
You'll see a table with your project's tasks:
Each task can be individually optimized for full granular control. Click on the task to access the optimization tools.
Step 5: Access Task Optimization Tools
Click the Actions button for your Review Summarizer task to access optimization capabilities:
Available Actions
Prompt Optimization: Fine-tune how your AI responds
Input Optimization: Improve input handling
Model Optimization: Test different AI models
Edge Case Detection: Find and fix problematic scenarios
Evaluations: Set success criteria
Task Information
You can also view task details including:
Task ID: Unique identifier for your task
Created: When the task was generated
Status: "Active" (ready to use)
Step 6: Add Test Inputs
Before optimizing, you need test data. Click Actions → Input Optimization:
Manual Inputs
Click "Add Input"
Add sample review text like:
"The product arrived quickly but the packaging was damaged. Overall satisfied with the purchase but disappointed in the delivery experience."Add several more examples with different sentiments and scenarios
What This Does
Provides data for optimization
Helps you see how your application performs
Creates a baseline for improvement
Step 7: Set Up Evaluations
Click Actions → Evaluations to define what success looks like:
Example Evaluation Criteria
"Correct Sequence": Information appears in logical order
"Accurate Details": Extracted details match the original text
"Complete Summary": All key points are captured
You can either:
Add manually: Write your own criteria
Generate automatically: Let Empromptu create evaluations for you
Step 8: Run Your First Optimization
Now you're ready to optimize! Here's the basic workflow:
Prompt Optimization
Click Actions → "Prompt Optimization"
You'll see tabs for:
Event Log: Shows all optimization attempts and scores
Prompt Family: Your collection of prompts
Manual Optimization: Step-by-step improvement wizard
Automatic Optimization: Let Empromptu optimize for you
Run Automatic Optimization
Navigate to the "Automatic Optimization" tab
Click the optimization button
Watch as Empromptu automatically improves your prompts
See your accuracy scores improve in real-time
Understanding Results
Scores: You'll see improvements like 5.5 → 7.0 → 8.5
Event Log: Shows each optimization attempt with detailed scoring
Prompt Family: New prompts are added to work together
Step 9: Deploy Your Application
Once you're satisfied with your optimization results, go back to the Builder interface to deploy:
Deployment Options
In the top right of the Builder interface, you'll see deployment buttons:
Deploy to Netlify: One-click deployment to Netlify hosting
Deploy to GitHub: Push your application to a GitHub repository
Download Project: Get files to host anywhere you want
Choose Your Deployment
Click your preferred deployment option
Follow the setup prompts (authentication, configuration)
Your Review Summarizer goes live within minutes
What You've Accomplished
🎉 Congratulations! You've successfully:
✅ Built an AI application using natural language
✅ Understood the project and task structure
✅ Added test inputs for optimization
✅ Set up evaluation criteria
✅ Ran optimization to improve performance
✅ Deployed your application live
Next Steps
Now that you've built your first application, explore more advanced features:
Model Optimization: Test different AI models for better performance
Edge Case Detection: Handle problematic scenarios
Advanced Prompt Families: Build sophisticated prompt collections
End User Analytics: Monitor real user interactions
Common Questions
Q: How long does optimization take? A: Automatic optimization usually completes within a few minutes. Manual optimization depends on how much you want to fine-tune.
Q: Can I modify the application after deployment? A: Yes! You can continue optimizing and redeploy at any time.
Q: What if my accuracy scores are low? A: Low scores (0-3) are normal for initial runs. Use the optimization tools to improve performance systematically.
Last updated