The agent will generate a comprehensive project description based on your initial description and any files you’ve shared. The description includes:
Project Goal: Your business objective and why you need this annotation.
Training data for an ML model.
Evaluating model performance.
Creating a benchmark dataset.
Quality assurance.
Annotation Task: The specific task that experts will perform.
What they will evaluate, classify, generate, or extract.
How they will interact with the data.
Context (if available from your description or files):
Data types and formats;
Label schema and categories;
Quality expectations.
What to check
Review the agent’s proposed description carefully. This is your chance to ensure the agent understands your task correctly. If the task is misunderstood here, the entire project will be misconfigured.
Check that:
✅ The Project Goal matches your business objective.
✅ The Annotation Task accurately describes what experts should do.
✅ All important aspects of your task are included.
✅ Any data types, formats, or categories mentioned are correct.
If something is wrong or missing:
Correct the agent—it will regenerate the description.
Make sure both you and the agent understand the task the same way.
Only proceed when you’re confident the description is accurate.