diff --git a/README.md b/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..814f36d8264c7350a87604acb5e525cbfd26fd3c
--- /dev/null
+++ b/README.md
@@ -0,0 +1,26 @@
+# Mixtral 8x7b and LlamaIndex
+
+This is a quick demo of how to get a fully local model up and running with Mixtral 8x7b and LlamaIndex; check out the accompanying [blog post](https://TKTKTK) for more details.
+
+## Install dependencies
+
+```
+poetry shell
+poetry install
+```
+
+## 1_smoketest.py
+
+After you've downloaded [Ollama](https://ollama.ai/) and run `ollama run mixtral` run this to make sure everything is connected.
+
+## 2_index_data.py
+
+This will index the data in the `data` folder. You can run this as many times as you want, and it will only index new data.
+
+## 3_verify_index.py
+
+This demonstrates the working index without re-loading the documents.
+
+## app.py
+
+A very simple Flask app that demonstrates how to use the index in a web app. Note that `flask run` will not work; you must run `python app.py`.