diff --git a/deployments/kubehound/ui/README.md b/deployments/kubehound/ui/README.md
new file mode 100644
index 000000000..dfa4de4d9
--- /dev/null
+++ b/deployments/kubehound/ui/README.md
@@ -0,0 +1,41 @@
+# UI
+
+## Jupyter Notebooks
+
+This UI is using Amazon's [Graph Notebook](https://github.com/aws/graph-notebook) tooling to query JanusGraph.
+
+As a first step, we recommend following the [KubeHound DSL 101 notebook](./KubehoundDSL_101.ipynb). You will find details on how to connect, default configuration and how to construct requests there.
+
+## Gremlin tricks
+
+### Cell magic
+
+As stated in the 101 notebook, to launch gremlin queries against the KubeHound graph, you need to start the cell with the jupyter cell magic `%%gremlin`
+
+### Display help
+
+This magic can take a bunch of parameters, you can get the help message by running a cell containing just
+```python
+%%gremlin --help
+g
+```
+
+### Common arguments
+
+Most of the pre-written cells in the notebooks will already have magic arguments defined:
+- `-d`: Displayed value on the node, defaults to the `label`, but often set to `name` or `class`
+- `-g`: Group by - when set (often to `critical`), groups the nodes to reduce visual clobber
+- `le 50`: Max labels length (default to 10, so overridden to 50 to fit the attacks names)
+- `-p inv,oute`: [Visualization hints](https://docs.aws.amazon.com/neptune/latest/userguide/notebooks-visualization.html#notebooks-visualization-Gremlin)
+
+### Save result
+
+You can save the output of the gremlin query to manipulate it in python with the `--store-to <var>` argument
+```python
+%%gremlin --store-to nsVertexCount
+g.V().groupCount().by('namespace')
+```
+And in a following cell
+```python
+print(nsVertexCount[0])
+```