Project and Tasks

Projects and Tasks are the core structure of Empromptu. Projects represent your AI applications, while Tasks handle specific functions within each application.

What you'll learn ⏱️ 4 minutes

• How projects organize your AI applications

• What the project dashboard shows you

• Understanding the tasks table and metrics

• How to navigate between projects and tasks

• Using the Actions button to access optimization tools

Understanding Projects

Projects are your complete AI applications. Each project represents one AI app that you've built using the Builder interface.

Examples of Projects:

• Customer Support Bot - Handles customer inquiries and support requests

• Review Summarizer - Analyzes customer feedback and extracts insights

• Content Generator - Creates marketing copy and blog posts

• Data Processor - Extracts and structures information from documents

Each project contains the complete AI application with model, RAG, and infrastructure included.

Project Dashboard Overview

When you click the Projects tab and open a project, you'll see the project dashboard:

Project Header

Review SummarizerA project that analyzes customer reviews and extracts key insights.

The header shows your project name and description (automatically generated from your Builder input).

Project Metrics

The dashboard displays key performance indicators:

0.00%                    0.00%                    2Average Initial Accuracy  Average Current Accuracy  Total TasksProgress: 0.00%

Metric explanations:

• Average Initial Accuracy: Performance when tasks first ran

• Average Current Accuracy: Current performance after optimization

• Total Tasks: Number of tasks within this project

• Progress: Overall optimization progress percentage

New Task Button

Click "+ New Task" to add additional tasks to your project. This is useful when you want to expand your AI application with new capabilities.

Understanding Tasks

Tasks are specific functions within your project. Each task handles a particular type of operation or workflow.

Task Examples within a Customer Support Project:

• Password Reset Task - Handles password reset requests

• Billing Inquiry Task - Processes billing questions

• Technical Support Task - Routes technical issues

• Refund Request Task - Manages refund processing

Tasks Table

The main part of your project dashboard is the tasks table:

Column Explanations:

Task Name: The identifier for this specific function

• Usually descriptive of what the task does

• Generated automatically but can be customized

Description: What this task accomplishes

• Automatically created based on your Builder input

• Can be edited to provide more clarity

Initial Accuracy: First performance measurement

• Shows "N/A" until the task runs with optimization

• Baseline score from 0-10 scale

Current Accuracy: Latest performance score

• Shows "N/A" until optimization begins

• Updates as you improve the task

Improvement: Progress made through optimization

• Shows "N/A" initially

• Displays positive/negative change from initial to current

Status: Current state of the task

• Active: Task is running and available for optimization

• Inactive: Task is paused or disabled

Actions: Gateway to optimization tools

• Click this button to access all optimization capabilities

• Most important feature for improving task performance

Using the Actions Button

The Actions button is your gateway to optimization tools. Click Actions on any task to access:

Available Actions:

• Prompt Optimization: Build and refine Prompt Families

• Input Optimization: Manage manual and end-user inputs

• Model Optimization: Test different AI models

• Edge Case Detection: Find and fix problematic scenarios

• Evaluations: Set up success criteria

When to Use Actions:

• Before deployment: Optimize performance for production

• After issues arise: Fix problems identified in real usage

• Regular maintenance: Continuously improve accuracy

• New requirements: Adapt to changing business needs

Task Status and Management

Task Status Types:

Active Tasks:

• Currently running and processing inputs

• Available for optimization and monitoring

• Can be deployed and used by end users

Managing Tasks:

• View task details: Click Actions to see comprehensive information

• Optimize performance: Use Actions → optimization tools

• Monitor metrics: Track accuracy improvements over time

• Add new tasks: Use "+ New Task" for additional functionality

Navigation Best Practices

Efficient Project Management:

1. Start with project overview: Check overall metrics first

2. Identify priority tasks: Focus on tasks with low accuracy scores

3. Use Actions strategically: Optimize high-impact tasks first

4. Monitor progress: Regular check project-level improvements

Workflow Recommendations:

• New projects: Set up evaluations before optimizing

• Existing projects: Monitor current accuracy for decline

• Multi-task projects: Optimize tasks individually, then review overall project performance

• Before deployment: Ensure all critical tasks meet accuracy requirements

Project Organization Tips

Naming Conventions:

• Projects: Use clear, descriptive names (e.g., "Customer Support Bot", "Invoice Processor")

• Tasks: Include action and scope (e.g., "Billing Inquiry Handler", "Product Review Analyzer")

Task Grouping:

• Keep related functions together: Group similar operations in one project

• Separate distinct applications: Different AI apps should be separate projects

• Consider user workflows: Tasks should match how users interact with your application

Common Project Patterns

Single-Task Projects:

Document Analyzer Project├── PDF Text Extraction Task

Simple applications with one main function.

Multi-Task Projects:

Customer Support Project  ├── Password Reset Task├── Billing Inquiry Task├── Technical Support Task└── Escalation Routing Task

Complex applications with multiple specialized functions.

Workflow Projects:

Content Creation Project├── Topic Research Task├── Draft Generation Task├── Review and Editing Task└── Publication Formatting Task

Applications that handle sequential processes.

Troubleshooting Common Issues

Tasks Showing N/A Metrics:

Cause: Task hasn't been optimized yet

Solution: Click Actions → run any optimization tool to generate initial metrics

Low Project-Level Accuracy:

Cause: Individual tasks need optimization

Solution: Focus on optimizing the lowest-performing tasks first

Can't Find Expected Tasks:

Cause: Tasks might be in different projects or need creation

Solution: Check project structure or use "+ New Task" to add missing functionality

Next Steps

Now that you understand Projects and Tasks:

• Learn Task Actions: Deep dive into the Actions button and optimization tools

• Understand Accuracy Scoring: Learn how the 0-10 scoring system works

• Explore Builder Interface: Create new projects and tasks effectively

• Start Optimizing: Begin improving your task performance