Complete User Guide

AIChemy Documentation

Everything you need to master prompt optimization and get the best results from AI.

Getting Started

Quick Start Guide

  1. 1

    Go to the Workspace

    Navigate to Dashboard → Workspace. This is where all the magic happens.

  2. 2

    Enter Your Prompt

    Type or paste your prompt into the input area. You'll see a real-time score appear.

  3. 3

    Select an Agent

    Choose the AI agent that best matches your task. Not sure? Start with "Comprehensive".

  4. 4

    Click "Improve Prompt"

    Watch as AIChemy transforms your prompt. See the before/after comparison and improvements.

  5. 5

    Copy and Use

    Copy your improved prompt and use it with any AI (ChatGPT, Claude, Gemini, etc.)

Pro Tip: Install the Chrome extension to improve prompts directly on ChatGPT, Claude, and other AI platforms without leaving the page!

AI Agents

AIChemy uses 10 specialized AI agents, each trained to improve a specific aspect of your prompts. Select the right agent for your task to get the best results.

Each agent applies different optimization techniques. The "Comprehensive" agent combines techniques from all agents for maximum improvement.

For Code & Technical Tasks

Use: Clarity, Specificity, or Comprehensive

These ensure precise requirements and clear output formats.

For Creative Writing

Use: Tone, Persona, or Context

These add style, character, and rich background.

For Math & Logic

Use: Chain of Thought

Forces step-by-step reasoning for accurate results.

For Formatted Outputs

Use: Few-Shot or Structure

Demonstrates exact output format with examples.

Workspace

The Workspace is your main hub for improving prompts. It provides a dual-panel interface showing your original prompt and the improved version side by side.

Input Panel

  • Type or paste your original prompt
  • Real-time character, word, and line count
  • Live score updates as you type
  • Syntax highlighting for code prompts

Improved Panel

  • See the enhanced prompt with improvements highlighted
  • View before/after score comparison
  • One-click copy to clipboard
  • Save to favorites for later use

Templates

Templates are pre-built prompt structures with customizable variables. They save time and ensure consistent, high-quality prompts for common tasks.

How Template Variables Work

Static Variables in Templates

Each template defines its variables like {{topic}} or {{language}}. These placeholders are static parts of the template structure.

Fill Variables with Your Content

When you click "Use", a form appears where you fill in each variable. Your values are automatically substituted everywhere the variable appears in the template.

Instant Substitution

If a variable like {{topic}} appears 3 times in the template, all 3 instances get replaced with your value (e.g., "global warming").

Example: Variable Substitution

Template (Before)
Write a blog post about {{topic}}.
Target audience: {{audience}}
Focus on {{topic}} benefits.
Result (After Filling)
Write a blog post about global warming.
Target audience: students
Focus on global warming benefits.

Notice how both {{topic}} placeholders were replaced with "global warming"

  1. 1
    Browse and select a template

    Go to Templates page and find one that fits your task.

  2. 2
    Click "Use" button

    A form opens showing all the template's variables.

  3. 3
    Fill in each variable

    Enter your values. Required fields are marked with *. See live preview as you type.

  4. 4
    Submit to workspace

    Your filled template loads in the workspace, ready for improvement.

  5. 5
    Apply the recommended agent

    The template's agent is pre-selected. Click "Improve" for best results.

Power User Tip: Variables are automatically extracted from your template content. When creating custom templates, just use {{variable_name}} syntax and the system will detect them!

AI Providers

AIChemy supports multiple AI providers. Connect your own API keys to use your preferred models, or use Google AI with OAuth for a seamless experience.

OpenAI

Industry-leading models with excellent general-purpose capabilities.

Models:GPT-5.1, GPT-4o
Best for:General tasks, coding, creative writing
Pricing:$5-60 per 1M tokens

Anthropic

Advanced reasoning with strong safety features and long context windows.

Models:Claude 3.5 Sonnet, Claude 3 Opus
Best for:Analysis, reasoning, safe outputs
Pricing:$0.80-75 per 1M tokens

Google AI

Multimodal capabilities with fast performance.

Models:Gemini 2.0 Pro, Gemini 1.5 Flash
Best for:Multimodal tasks, fast responses
Pricing:$1.25-10 per 1M tokens

Mistral AI

European AI with excellent performance-to-cost ratio.

Models:Mistral Large, Mistral Medium
Best for:Code generation, European compliance
Pricing:$2-8 per 1M tokens

Groq

Ultra-fast inference with dedicated LPU hardware.

Models:Llama 3.1 70B, Mixtral 8x7B
Best for:Speed-critical applications
Pricing:Free tier available

Meta (Llama)

Open-source models with strong community support.

Models:Llama 3.1 405B, Llama 3.1 70B
Best for:Open-source requirements, customization
Pricing:Varies by provider
  1. 1
    Go to Settings → AI Providers

    Find the provider you want to enable.

  2. 2
    Enter your API key

    Get your API key from the provider's website (links provided).

  3. 3
    Toggle the provider ON

    Enable the provider and its models will appear in the workspace.

Google AI Special: You can connect Google AI using OAuth (Google Sign-In) instead of an API key for easier setup.

Settings

Customize AIChemy to match your workflow. All settings are stored securely and synced across devices.

  • Enable/disable individual providers
  • Enter and manage API keys securely
  • Set default models per provider
  • Connect Google AI via OAuth
  • Configure settings per domain (Coding, Creative, etc.)
  • Set default agent for each segment
  • Adjust temperature per segment
  • Add custom instructions for each domain

Temperature Guide

Temperature controls how "creative" or "random" the AI's responses are. Lower values produce more consistent, focused outputs. Higher values yield more varied, creative results.

Temperature Scale

0.0Deterministic
0.3Focused
0.5Balanced
0.7Creative
1.0Maximum

Recommended Temperatures by Task

Programming & Code

Default Agent: Comprehensive
0.3

Mathematics

Default Agent: Chain of Thought
0.2

Science & Research

Default Agent: Expert
0.4

Languages & Writing

Default Agent: Tone
0.7

Business & Professional

Default Agent: Specificity
0.5

Creative & Art

Default Agent: Persona
0.9
Important: For tasks requiring accuracy (math, coding, facts), use lower temperatures (0.2-0.4). For creative tasks, higher temperatures (0.7-0.9) allow more exploration.

Prompt Scoring

AIChemy analyzes your prompts in real-time and provides a score from 0-100 based on five key dimensions.

Clarity

20%

How clear and unambiguous

Specificity

25%

Level of detail and precision

Context

20%

Background information provided

Structure

15%

Organization and format

Effectiveness

20%

Action verbs and output specs

Score Interpretation

0-40

Needs significant improvement

41-60

Fair, can be improved

61-80

Good quality prompt

81-100

Excellent, optimized

History & Favorites

AIChemy automatically saves your improvements so you can revisit and reuse your best prompts.

History Features

  • All improvements saved with timestamps
  • Original and improved versions stored
  • Filter by date, segment, or search
  • Bulk delete and management tools

Favorites System

  • Star prompts you want to keep
  • Quick filter to view favorites only
  • One-click copy to reuse
  • Perfect for frequently used prompts

Pro Tips

Use Domain Segments: Setting the right domain (Coding, Creative, etc.) automatically adjusts temperature and selects the best default agent for your task type.
Chain Improvements: You can run multiple agents on the same prompt sequentially. Start with Structure, then apply Specificity for highly organized, detailed prompts.
Save Custom Instructions: In Settings → Domain Segments, add custom instructions that will be included in every improvement. Great for brand voice or specific requirements.
Use the Extension: The Chrome extension lets you improve prompts directly on ChatGPT, Claude, Gemini, and other AI platforms without copy-pasting.
Check the Score Breakdown: After improving, look at individual metric scores. If one area is still low, try an agent that specifically addresses that dimension.
Model Selection Matters: Different models have different strengths. Use GPT-5.1 or Claude for complex reasoning, Gemini for multimodal tasks, and Groq for ultra-fast responses.

Need More Help?

Have questions or feedback? We'd love to hear from you. Reach out via email and we'll get back to you as soon as possible.