Product Guide
How PromptLab Works
PromptLab helps you turn prompt writing into a real workflow. Create prompt versions, optimize them with AI, compare outputs across models, refine the best result, and track what wins over time.
The Workflow
Step-by-Step Guide
Everything you can do in PromptLab
Create a Prompt
Start by creating a new prompt. A prompt is your main workspace for one task, one idea, or one use case. Think of it as the home base for everything you want to test and improve.
Examples: Product Description Generator, SAT Tutor Prompt, Sales Outreach Prompt, Customer Support Reply Prompt.
Add or Improve a Version
Inside each prompt, create your first version — the actual prompt text you want to test. You can create multiple versions to try different wording, structure, tone, or strategy.
This helps you move from rough ideas to stronger prompt candidates without losing your originals.
Add a Test Input
Enter the example request, task, question, or scenario you want to test. This is the input the AI responds to when you run a comparison.
Examples: Write an eBay description, explain a math problem, write a cold outreach email, summarize a customer complaint.
You can also save useful test inputs and reuse them later.
Choose How to Compare
PromptLab gives you two ways to compare results.
Compare Versions
Test multiple prompt versions against the same input to see which wording performs better.
Compare Models
Test the same prompt version across different AI models like GPT, Claude, Gemini, and Grok.
This helps you understand whether the improvement comes from the prompt, the model, or both.
Review Results Side by Side
After running a test, PromptLab shows the outputs in a clean comparison view so you can quickly evaluate differences.
Instead of guessing across separate chat windows, you can compare everything in one organized place.
Pick the Winner
When you find the strongest result, mark it as the winner. This gives your test a clear decision point and helps you keep track of what performed best.
PromptLab helps you turn experiments into actual decisions.
Continue the Best Result
If the best result is close but not perfect, continue the conversation with that specific result and refine it further.
Use follow-up conversation to improve clarity, change tone, shorten or expand the output, or push it closer to exactly what you want.
Track What Wins
Use Race Stats to see which models win most often across your model-comparison runs.
This helps you spot patterns in quality, speed, cost, and overall performance across your workflows.
Save and Reuse What Works
As you test and improve, PromptLab helps you keep what matters. Save stronger prompt versions, reusable inputs, successful runs, and refined outputs so you can build on what already works.
Over time, this turns PromptLab into a real working system instead of a collection of random experiments.
Prompt writing should be a process, not guesswork.
Create a prompt, improve it, compare outputs, refine the best result, track what wins, and save what works. That is how better prompts are built.
Ready to test smarter?
Start your first prompt and see what works faster.