diff --git a/docs/examples/client_example.ipynb b/docs/examples/client_example.ipynb
index d45f1fb..d7cb2ba 100644
--- a/docs/examples/client_example.ipynb
+++ b/docs/examples/client_example.ipynb
@@ -1497,6 +1497,120 @@
     "client.get_item(item_id=\"NMC811_e1_c1\", load_blocks=True, display=True);"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "id": "0683d0d5-c0b9-43bc-b02b-9b1add502c1e",
+   "metadata": {},
+   "source": [
+    "## Exploring relationships between items"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "241be170-44e3-4ff5-a477-f7c992379fb7",
+   "metadata": {},
+   "source": [
+    "Items in *datalab* can be linked together, e.g., a cell may be made from several samples, which in turn were synthesised from starting materials.\n",
+    "The graph of these relationships can be retrieved with `get_item_graph`, either for all items you can see, or centred on a particular item.\n",
+    "\n",
+    "With `as_networkx=True`, the graph is returned as a [`networkx.DiGraph`](https://networkx.org/documentation/stable/reference/classes/digraph.html), with edges pointing from parent to child.\n",
+    "This requires the optional `networkx` dependency (`pip install datalab-api[networkx]`); we will also use `matplotlib` to draw the graph below."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "10918dcc-53bf-497f-a40f-7b78d84e03f6",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "graph = client.get_item_graph(item_id=\"NMC811_e1\", max_depth=2, as_networkx=True)\n",
+    "print(f\"{graph.number_of_nodes()} items, {graph.number_of_edges()} relationships\")"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "81e1c738-74d3-4c1a-ada2-f023c23f98b2",
+   "metadata": {},
+   "source": [
+    "Each node is keyed by its item ID, with the rest of the item summary (e.g., `name` and `type`) stored as node attributes:"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "b86b0122-2a5d-4242-b8a5-7d6118de74db",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "for item_id, data in graph.nodes(data=True):\n",
+    "    print(f\"{item_id:<30} {data.get('type', ''):<20} {data.get('name', '')}\")"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "33102cef-4b72-46d0-8576-c4cd21b46b03",
+   "metadata": {},
+   "source": [
+    "The graph can then be analysed or visualised with any of the usual `networkx` tools, for example, colouring each item by its type:"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "2dd36755-0ac3-4711-8820-230c0242a147",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import matplotlib.pyplot as plt\n",
+    "import networkx as nx\n",
+    "\n",
+    "type_colours = {\n",
+    "    \"samples\": \"tab:blue\",\n",
+    "    \"starting_materials\": \"tab:green\",\n",
+    "    \"cells\": \"tab:orange\",\n",
+    "    \"equipment\": \"tab:purple\",\n",
+    "}\n",
+    "node_colours = [\n",
+    "    type_colours.get(data.get(\"type\"), \"tab:gray\") for _, data in graph.nodes(data=True)\n",
+    "]\n",
+    "\n",
+    "fig, ax = plt.subplots(figsize=(10, 8))\n",
+    "pos = nx.spring_layout(graph, seed=42)\n",
+    "nx.draw_networkx(\n",
+    "    graph,\n",
+    "    pos=pos,\n",
+    "    ax=ax,\n",
+    "    node_color=node_colours,\n",
+    "    node_size=800,\n",
+    "    font_size=8,\n",
+    "    arrowsize=15,\n",
+    ")\n",
+    "for item_type, colour in type_colours.items():\n",
+    "    ax.scatter([], [], color=colour, label=item_type)\n",
+    "ax.legend(loc=\"best\")\n",
+    "ax.axis(\"off\")\n",
+    "plt.show()"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "a10a9032-8299-43c9-9eca-79bc4421fe18",
+   "metadata": {},
+   "source": [
+    "Standard graph algorithms can also be used to answer questions about your items, for example, finding everything that went into making a given item:"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "04e9f79b-8c07-4778-b2ed-dd82aa735cc4",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "nx.ancestors(graph, \"NMC811_e1\")"
+   ]
+  },
   {
    "cell_type": "markdown",
    "id": "243e941b-c465-4395-94f8-e8db441aece2",