diff --git a/docs/usage/retrieval_core.md b/docs/usage/retrieval_core.md
new file mode 100644
index 0000000..8065e5e
--- /dev/null
+++ b/docs/usage/retrieval_core.md
@@ -0,0 +1,14 @@
+# Retrieval Core Task Usage
+## What it does
+The `retrieval_core` task acts like a mini search engine. It generates a bunch of fake text documents and then calculates how many times specific words appear across those documents. 
+Instead of checking every single file, it uses math to sample just a few of them to estimate the final results. This is great for testing probabilistic search approximations.
+## Example Script
+Here's how you can run the task in Python to generate some fake data and solve it:
+```python
+from chuck.tasks import retrieval_core
+# 1. Generate 100 fake documents
+payload = retrieval_core.generate(count=100, seed=42)
+# 2. Run the task to search through them
+solution = retrieval_core.solve(payload)
+# 3. Check the results
+print(solution)
\ No newline at end of file