diff --git a/docs/examples/scibert/data.ipynb b/docs/examples/scibert/data.ipynb
index 4487b65..f1a6b4a 100644
--- a/docs/examples/scibert/data.ipynb
+++ b/docs/examples/scibert/data.ipynb
@@ -26,56 +26,8 @@
     "\n",
     "Use Python 3.12 or 3.13 and install this checkout with `uv pip install --editable '.[notebooks]'`.\n",
     "See **Run notebooks locally** in the repository README for the kernel, execution order,\n",
-    "and full-data options. Generated artifacts stay under `.cache/notebooks/`.\n"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 1,
-   "id": "4bf6cce1",
-   "metadata": {},
-   "outputs": [
-    {
-     "name": "stderr",
-     "output_type": "stream",
-     "text": [
-      "\u001b[38;5;226mEnvironment variable ENVIRONMENT is not set, defaulting to development mode ‼️\u001b[0m\n",
-      "\u001b[38;5;226mThe selected tracing database (ParallelTinyDbDriver) is not recommended for production\u001b[0m\n",
-      "\u001b[38;5;39mGreatAI (v0.1.11): configured ✅\u001b[0m\n",
-      "\u001b[38;5;39m  🔩 tracing_database: ParallelTinyDbDriver\u001b[0m\n",
-      "\u001b[38;5;39m  🔩 large_file_implementation: LargeFileLocal\u001b[0m\n",
-      "\u001b[38;5;39m  🔩 is_production: False\u001b[0m\n",
-      "\u001b[38;5;39m  🔩 should_log_exception_stack: True\u001b[0m\n",
-      "\u001b[38;5;39m  🔩 prediction_cache_size: 512\u001b[0m\n",
-      "\u001b[38;5;39m  🔩 dashboard_table_size: 50\u001b[0m\n"
-     ]
-    },
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "Notebook artifacts: /Users/schmelczer/Projects/great-ai/.cache/notebooks/scibert\n"
-     ]
-    }
-   ],
-   "source": [
-    "import os\n",
-    "import sys\n",
-    "from pathlib import Path\n",
-    "\n",
-    "# Works when opened from the repository root or any notebook directory.\n",
-    "_start = Path(globals().get(\"__file__\", Path.cwd())).resolve()\n",
-    "ROOT = next(\n",
-    "    p\n",
-    "    for p in [_start, *_start.parents]\n",
-    "    if (p / \"pyproject.toml\").is_file() and (p / \"great_ai\").is_dir()\n",
-    ")\n",
-    "sys.path.insert(0, str(ROOT / \"docs\"))\n",
-    "from notebook_support import setup_notebook, mag_lines\n",
-    "\n",
-    "FULL_RUN = os.environ.get(\"GREAT_AI_NOTEBOOK_FULL\") == \"1\"\n",
-    "WORK_DIR = setup_notebook(ROOT, \"scibert\")\n",
-    "DATA_DIR = ROOT / \"docs/tutorial/data\""
+    "and full-data options. Run related notebooks from the same working directory\n",
+    "to share the default `.cache/` model store and `tracing_database.json` database.\n"
    ]
   },
   {
@@ -94,7 +46,10 @@
    ],
    "source": [
     "import json\n",
-    "from notebook_support import demo_annotations\n",
+    "import os\n",
+    "from pathlib import Path\n",
+    "\n",
+    "from docs.notebook_support import demo_annotations\n",
     "\n",
     "annotation_dir = os.environ.get(\"GREAT_AI_ANNOTATIONS_DIR\")\n",
     "if annotation_dir:\n",
diff --git a/docs/examples/scibert/deploy.ipynb b/docs/examples/scibert/deploy.ipynb
index dd0f0b1..2467494 100644
--- a/docs/examples/scibert/deploy.ipynb
+++ b/docs/examples/scibert/deploy.ipynb
@@ -22,68 +22,8 @@
     "\n",
     "Use Python 3.12 or 3.13 and install this checkout with `uv pip install --editable '.[notebooks]'`.\n",
     "See **Run notebooks locally** in the repository README for the kernel, execution order,\n",
-    "and full-data options. Generated artifacts stay under `.cache/notebooks/`.\n"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 9,
-   "id": "3775e11d",
-   "metadata": {},
-   "outputs": [
-    {
-     "name": "stderr",
-     "output_type": "stream",
-     "text": [
-      "\u001b[38;5;196mConfiguration has been already initialised, overwriting.\n",
-      "Make sure to call `configure()` before importing your application code.\u001b[0m\n",
-      "\u001b[38;5;226mEnvironment variable ENVIRONMENT is not set, defaulting to development mode ‼️\u001b[0m\n",
-      "\u001b[38;5;226mThe selected tracing database (ParallelTinyDbDriver) is not recommended for production\u001b[0m\n",
-      "\u001b[38;5;39mGreatAI (v0.1.11): configured ✅\u001b[0m\n",
-      "\u001b[38;5;39m  🔩 tracing_database: ParallelTinyDbDriver\u001b[0m\n",
-      "\u001b[38;5;39m  🔩 large_file_implementation: LargeFileLocal\u001b[0m\n",
-      "\u001b[38;5;39m  🔩 is_production: False\u001b[0m\n",
-      "\u001b[38;5;39m  🔩 should_log_exception_stack: True\u001b[0m\n",
-      "\u001b[38;5;39m  🔩 prediction_cache_size: 512\u001b[0m\n",
-      "\u001b[38;5;39m  🔩 dashboard_table_size: 50\u001b[0m\n"
-     ]
-    },
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "Notebook artifacts: /Users/schmelczer/Projects/great-ai/.cache/notebooks/scibert\n"
-     ]
-    }
-   ],
-   "source": [
-    "import os\n",
-    "import sys\n",
-    "from pathlib import Path\n",
-    "\n",
-    "# Works when opened from the repository root or any notebook directory.\n",
-    "_start = Path(globals().get(\"__file__\", Path.cwd())).resolve()\n",
-    "ROOT = next(\n",
-    "    p\n",
-    "    for p in [_start, *_start.parents]\n",
-    "    if (p / \"pyproject.toml\").is_file() and (p / \"great_ai\").is_dir()\n",
-    ")\n",
-    "sys.path.insert(0, str(ROOT / \"docs\"))\n",
-    "from notebook_support import setup_notebook, mag_lines\n",
-    "\n",
-    "FULL_RUN = os.environ.get(\"GREAT_AI_NOTEBOOK_FULL\") == \"1\"\n",
-    "WORK_DIR = setup_notebook(ROOT, \"scibert\")\n",
-    "DATA_DIR = ROOT / \"docs/tutorial/data\""
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 10,
-   "id": "cell-3",
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "# The setup cell selects local model storage and a local TinyDB tracing database."
+    "and full-data options. Run related notebooks from the same working directory\n",
+    "to share the default `.cache/` model store and `tracing_database.json` database.\n"
    ]
   },
   {
@@ -144,6 +84,7 @@
    ],
    "source": [
     "from functools import lru_cache\n",
+    "from pathlib import Path\n",
     "from great_ai import use_model\n",
     "from transformers import AutoModelForSequenceClassification, AutoTokenizer\n",
     "\n",
diff --git a/docs/examples/scibert/train.ipynb b/docs/examples/scibert/train.ipynb
index a39fad3..21378b0 100644
--- a/docs/examples/scibert/train.ipynb
+++ b/docs/examples/scibert/train.ipynb
@@ -27,56 +27,18 @@
     "\n",
     "Use Python 3.12 or 3.13 and install this checkout with `uv pip install --editable '.[notebooks]'`.\n",
     "See **Run notebooks locally** in the repository README for the kernel, execution order,\n",
-    "and full-data options. Generated artifacts stay under `.cache/notebooks/`.\n"
+    "and full-data options. Run related notebooks from the same working directory\n",
+    "to share the default `.cache/` model store and `tracing_database.json` database.\n"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 1,
+   "execution_count": null,
    "id": "824837fa",
    "metadata": {},
-   "outputs": [
-    {
-     "name": "stderr",
-     "output_type": "stream",
-     "text": [
-      "\u001b[38;5;226mEnvironment variable ENVIRONMENT is not set, defaulting to development mode ‼️\u001b[0m\n",
-      "\u001b[38;5;226mThe selected tracing database (ParallelTinyDbDriver) is not recommended for production\u001b[0m\n",
-      "\u001b[38;5;39mGreatAI (v0.1.11): configured ✅\u001b[0m\n",
-      "\u001b[38;5;39m  🔩 tracing_database: ParallelTinyDbDriver\u001b[0m\n",
-      "\u001b[38;5;39m  🔩 large_file_implementation: LargeFileLocal\u001b[0m\n",
-      "\u001b[38;5;39m  🔩 is_production: False\u001b[0m\n",
-      "\u001b[38;5;39m  🔩 should_log_exception_stack: True\u001b[0m\n",
-      "\u001b[38;5;39m  🔩 prediction_cache_size: 512\u001b[0m\n",
-      "\u001b[38;5;39m  🔩 dashboard_table_size: 50\u001b[0m\n"
-     ]
-    },
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "Notebook artifacts: /Users/schmelczer/Projects/great-ai/.cache/notebooks/scibert\n"
-     ]
-    }
-   ],
+   "outputs": [],
    "source": [
-    "import os\n",
-    "import sys\n",
-    "from pathlib import Path\n",
-    "\n",
-    "# Works when opened from the repository root or any notebook directory.\n",
-    "_start = Path(globals().get(\"__file__\", Path.cwd())).resolve()\n",
-    "ROOT = next(\n",
-    "    p\n",
-    "    for p in [_start, *_start.parents]\n",
-    "    if (p / \"pyproject.toml\").is_file() and (p / \"great_ai\").is_dir()\n",
-    ")\n",
-    "sys.path.insert(0, str(ROOT / \"docs\"))\n",
-    "from notebook_support import setup_notebook, mag_lines\n",
-    "\n",
-    "FULL_RUN = os.environ.get(\"GREAT_AI_NOTEBOOK_FULL\") == \"1\"\n",
-    "WORK_DIR = setup_notebook(ROOT, \"scibert\")\n",
-    "DATA_DIR = ROOT / \"docs/tutorial/data\""
+    "from docs.notebook_support import FULL_RUN"
    ]
   },
   {
@@ -244,7 +206,7 @@
     "    torch.set_num_threads(2)\n",
     "    vocabulary = [\"[PAD]\", \"[UNK]\", \"[CLS]\", \"[SEP]\", \"[MASK]\"]\n",
     "    vocabulary += sorted(set(re.findall(r\"[a-z]+|[0-9]+|[^\\w\\s]\", \" \".join(X).lower())))\n",
-    "    vocab_path = WORK_DIR / \"vocab.txt\"\n",
+    "    vocab_path = Path(\"vocab.txt\")\n",
     "    vocab_path.write_text(\"\\n\".join(vocabulary) + \"\\n\", encoding=\"utf-8\")\n",
     "    tokenizer = BertTokenizer(vocab=str(vocab_path), do_lower_case=True)\n",
     "    model = BertForSequenceClassification(\n",
@@ -288,7 +250,7 @@
     "\n",
     "\n",
     "training_args = TrainingArguments(\n",
-    "    output_dir=WORK_DIR / \"models\",\n",
+    "    output_dir=\"models\",\n",
     "    per_device_train_batch_size=BATCH_SIZE,\n",
     "    per_device_eval_batch_size=BATCH_SIZE,\n",
     "    save_total_limit=5,\n",
@@ -321,7 +283,7 @@
     "# This evaluates the selected checkpoint on validation, not an untouched test set.\n",
     "validation_metrics = trainer.evaluate(metric_key_prefix=\"validation\")\n",
     "trainer.save_metrics(\"validation\", validation_metrics)\n",
-    "(WORK_DIR / \"models/validation-run.json\").write_text(\n",
+    "Path(\"models/validation-run.json\").write_text(\n",
     "    json.dumps(\n",
     "        {\n",
     "            \"mode\": \"scibert\" if FULL_RUN else \"tiny-random-bert-demo\",\n",
@@ -399,9 +361,9 @@
     "\n",
     "# Trainer has loaded the selected validation checkpoint. Save its tokenizer\n",
     "# too, so deployment never needs to download a different vocabulary.\n",
-    "trainer.save_model(str(WORK_DIR / \"pretrained\"))\n",
-    "tokenizer.save_pretrained(WORK_DIR / \"pretrained\")\n",
-    "save_model(WORK_DIR / \"pretrained\", key=\"scibert-highlights\", keep_last_n=2)"
+    "trainer.save_model(\"pretrained\")\n",
+    "tokenizer.save_pretrained(\"pretrained\")\n",
+    "save_model(Path(\"pretrained\"), key=\"scibert-highlights\", keep_last_n=2)"
    ]
   },
   {
diff --git a/docs/examples/simple-mag/train.ipynb b/docs/examples/simple-mag/train.ipynb
index dcee263..173717a 100644
--- a/docs/examples/simple-mag/train.ipynb
+++ b/docs/examples/simple-mag/train.ipynb
@@ -10,8 +10,8 @@
     "By default this notebook uses 100 training and 30 test samples per class\n",
     "and a small Naive Bayes parameter search for a quick local run. Set\n",
     "`GREAT_AI_NOTEBOOK_FULL=1` before launching Jupyter to use all bundled data\n",
-    "and the larger search. Generated figures are saved under the printed\n",
-    "artifact directory. Local subset scores do not reproduce the historical\n",
+    "and the larger search. Generated figures are saved in the working directory.\n",
+    "Local subset scores do not reproduce the historical\n",
     "full-dataset comparison with the 2019 SciBERT baseline (macro F1 0.6571)."
    ]
   },
@@ -24,56 +24,18 @@
     "\n",
     "Use Python 3.12 or 3.13 and install this checkout with `uv pip install --editable '.[notebooks]'`.\n",
     "See **Run notebooks locally** in the repository README for the kernel, execution order,\n",
-    "and full-data options. Generated artifacts stay under `.cache/notebooks/`.\n"
+    "and full-data options. Run related notebooks from the same working directory\n",
+    "to share the default `.cache/` model store and `tracing_database.json` database.\n"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 1,
+   "execution_count": null,
    "id": "4a11c71a",
    "metadata": {},
-   "outputs": [
-    {
-     "name": "stderr",
-     "output_type": "stream",
-     "text": [
-      "\u001b[38;5;226mEnvironment variable ENVIRONMENT is not set, defaulting to development mode ‼️\u001b[0m\n",
-      "\u001b[38;5;226mThe selected tracing database (ParallelTinyDbDriver) is not recommended for production\u001b[0m\n",
-      "\u001b[38;5;39mGreatAI (v0.1.11): configured ✅\u001b[0m\n",
-      "\u001b[38;5;39m  🔩 tracing_database: ParallelTinyDbDriver\u001b[0m\n",
-      "\u001b[38;5;39m  🔩 large_file_implementation: LargeFileLocal\u001b[0m\n",
-      "\u001b[38;5;39m  🔩 is_production: False\u001b[0m\n",
-      "\u001b[38;5;39m  🔩 should_log_exception_stack: True\u001b[0m\n",
-      "\u001b[38;5;39m  🔩 prediction_cache_size: 512\u001b[0m\n",
-      "\u001b[38;5;39m  🔩 dashboard_table_size: 50\u001b[0m\n"
-     ]
-    },
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "Notebook artifacts: /Users/schmelczer/Projects/great-ai/.cache/notebooks/simple-mag\n"
-     ]
-    }
-   ],
+   "outputs": [],
    "source": [
-    "import os\n",
-    "import sys\n",
-    "from pathlib import Path\n",
-    "\n",
-    "# Works when opened from the repository root or any notebook directory.\n",
-    "_start = Path(globals().get(\"__file__\", Path.cwd())).resolve()\n",
-    "ROOT = next(\n",
-    "    p\n",
-    "    for p in [_start, *_start.parents]\n",
-    "    if (p / \"pyproject.toml\").is_file() and (p / \"great_ai\").is_dir()\n",
-    ")\n",
-    "sys.path.insert(0, str(ROOT / \"docs\"))\n",
-    "from notebook_support import setup_notebook, mag_lines\n",
-    "\n",
-    "FULL_RUN = os.environ.get(\"GREAT_AI_NOTEBOOK_FULL\") == \"1\"\n",
-    "WORK_DIR = setup_notebook(ROOT, \"simple-mag\")\n",
-    "DATA_DIR = ROOT / \"docs/tutorial/data\""
+    "from docs.notebook_support import DATA_DIR, FULL_RUN, mag_lines"
    ]
   },
   {
@@ -149,7 +111,7 @@
     "ax.bar(domains, counts)\n",
     "ax.set_ylabel(\"Count\")\n",
     "fig.tight_layout()\n",
-    "fig.savefig(WORK_DIR / \"mag-distribution.png\", dpi=500)\n",
+    "fig.savefig(\"mag-distribution.png\", dpi=500)\n",
     "None"
    ]
   },
@@ -2806,7 +2768,7 @@
     "    values_format=\".2f\",\n",
     ")\n",
     "plt.tight_layout()\n",
-    "plt.savefig(WORK_DIR / \"mag-confusion.png\", dpi=600)"
+    "plt.savefig(\"mag-confusion.png\", dpi=600)"
    ]
   }
  ],
diff --git a/docs/examples/simple/data.ipynb b/docs/examples/simple/data.ipynb
index 4d0b397..500ad9a 100644
--- a/docs/examples/simple/data.ipynb
+++ b/docs/examples/simple/data.ipynb
@@ -29,56 +29,18 @@
     "\n",
     "Use Python 3.12 or 3.13 and install this checkout with `uv pip install --editable '.[notebooks]'`.\n",
     "See **Run notebooks locally** in the repository README for the kernel, execution order,\n",
-    "and full-data options. Generated artifacts stay under `.cache/notebooks/`.\n"
+    "and full-data options. Run related notebooks from the same working directory\n",
+    "to share the default `.cache/` model store and `tracing_database.json` database.\n"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 1,
+   "execution_count": null,
    "id": "a800dc4d",
    "metadata": {},
-   "outputs": [
-    {
-     "name": "stderr",
-     "output_type": "stream",
-     "text": [
-      "\u001b[38;5;226mEnvironment variable ENVIRONMENT is not set, defaulting to development mode ‼️\u001b[0m\n",
-      "\u001b[38;5;226mThe selected tracing database (ParallelTinyDbDriver) is not recommended for production\u001b[0m\n",
-      "\u001b[38;5;39mGreatAI (v0.1.11): configured ✅\u001b[0m\n",
-      "\u001b[38;5;39m  🔩 tracing_database: ParallelTinyDbDriver\u001b[0m\n",
-      "\u001b[38;5;39m  🔩 large_file_implementation: LargeFileLocal\u001b[0m\n",
-      "\u001b[38;5;39m  🔩 is_production: False\u001b[0m\n",
-      "\u001b[38;5;39m  🔩 should_log_exception_stack: True\u001b[0m\n",
-      "\u001b[38;5;39m  🔩 prediction_cache_size: 512\u001b[0m\n",
-      "\u001b[38;5;39m  🔩 dashboard_table_size: 50\u001b[0m\n"
-     ]
-    },
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "Notebook artifacts: /Users/schmelczer/Projects/great-ai/.cache/notebooks/simple\n"
-     ]
-    }
-   ],
+   "outputs": [],
    "source": [
-    "import os\n",
-    "import sys\n",
-    "from pathlib import Path\n",
-    "\n",
-    "# Works when opened from the repository root or any notebook directory.\n",
-    "_start = Path(globals().get(\"__file__\", Path.cwd())).resolve()\n",
-    "ROOT = next(\n",
-    "    p\n",
-    "    for p in [_start, *_start.parents]\n",
-    "    if (p / \"pyproject.toml\").is_file() and (p / \"great_ai\").is_dir()\n",
-    ")\n",
-    "sys.path.insert(0, str(ROOT / \"docs\"))\n",
-    "from notebook_support import setup_notebook, mag_lines\n",
-    "\n",
-    "FULL_RUN = os.environ.get(\"GREAT_AI_NOTEBOOK_FULL\") == \"1\"\n",
-    "WORK_DIR = setup_notebook(ROOT, \"simple\")\n",
-    "DATA_DIR = ROOT / \"docs/tutorial/data\""
+    "from docs.notebook_support import DATA_DIR, FULL_RUN, mag_lines"
    ]
   },
   {
@@ -157,10 +119,10 @@
    "source": [
     "### Load\n",
     "\n",
-    "Store the dataset in the local TinyDB tracing database. All notebooks in\n",
-    "this example share `.cache/notebooks/simple/` (or `simple-full/`). No S3 or\n",
-    "MongoDB credentials are needed. A dataset tag makes rerunning this cell\n",
-    "replace just this example's ground truth instead of duplicating it."
+    "Store the dataset in the default local TinyDB tracing database,\n",
+    "`tracing_database.json`. Run the data, training and deployment notebooks\n",
+    "from the same working directory to share it. A dataset tag makes rerunning\n",
+    "this cell replace just this example's ground truth instead of duplicating it."
    ]
   },
   {
diff --git a/docs/examples/simple/deploy.ipynb b/docs/examples/simple/deploy.ipynb
index 5981f57..ab6af0e 100644
--- a/docs/examples/simple/deploy.ipynb
+++ b/docs/examples/simple/deploy.ipynb
@@ -28,56 +28,8 @@
     "\n",
     "Use Python 3.12 or 3.13 and install this checkout with `uv pip install --editable '.[notebooks]'`.\n",
     "See **Run notebooks locally** in the repository README for the kernel, execution order,\n",
-    "and full-data options. Generated artifacts stay under `.cache/notebooks/`.\n"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 1,
-   "id": "b7a9d250",
-   "metadata": {},
-   "outputs": [
-    {
-     "name": "stderr",
-     "output_type": "stream",
-     "text": [
-      "\u001b[38;5;226mEnvironment variable ENVIRONMENT is not set, defaulting to development mode ‼️\u001b[0m\n",
-      "\u001b[38;5;226mThe selected tracing database (ParallelTinyDbDriver) is not recommended for production\u001b[0m\n",
-      "\u001b[38;5;39mGreatAI (v0.1.11): configured ✅\u001b[0m\n",
-      "\u001b[38;5;39m  🔩 tracing_database: ParallelTinyDbDriver\u001b[0m\n",
-      "\u001b[38;5;39m  🔩 large_file_implementation: LargeFileLocal\u001b[0m\n",
-      "\u001b[38;5;39m  🔩 is_production: False\u001b[0m\n",
-      "\u001b[38;5;39m  🔩 should_log_exception_stack: True\u001b[0m\n",
-      "\u001b[38;5;39m  🔩 prediction_cache_size: 512\u001b[0m\n",
-      "\u001b[38;5;39m  🔩 dashboard_table_size: 50\u001b[0m\n"
-     ]
-    },
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "Notebook artifacts: /Users/schmelczer/Projects/great-ai/.cache/notebooks/simple\n"
-     ]
-    }
-   ],
-   "source": [
-    "import os\n",
-    "import sys\n",
-    "from pathlib import Path\n",
-    "\n",
-    "# Works when opened from the repository root or any notebook directory.\n",
-    "_start = Path(globals().get(\"__file__\", Path.cwd())).resolve()\n",
-    "ROOT = next(\n",
-    "    p\n",
-    "    for p in [_start, *_start.parents]\n",
-    "    if (p / \"pyproject.toml\").is_file() and (p / \"great_ai\").is_dir()\n",
-    ")\n",
-    "sys.path.insert(0, str(ROOT / \"docs\"))\n",
-    "from notebook_support import setup_notebook, mag_lines\n",
-    "\n",
-    "FULL_RUN = os.environ.get(\"GREAT_AI_NOTEBOOK_FULL\") == \"1\"\n",
-    "WORK_DIR = setup_notebook(ROOT, \"simple\")\n",
-    "DATA_DIR = ROOT / \"docs/tutorial/data\""
+    "and full-data options. Run related notebooks from the same working directory\n",
+    "to share the default `.cache/` model store and `tracing_database.json` database.\n"
    ]
   },
   {
diff --git a/docs/examples/simple/stacked-bars.ipynb b/docs/examples/simple/stacked-bars.ipynb
index 2881941..c1a6545 100644
--- a/docs/examples/simple/stacked-bars.ipynb
+++ b/docs/examples/simple/stacked-bars.ipynb
@@ -9,56 +9,8 @@
     "\n",
     "Use Python 3.12 or 3.13 and install this checkout with `uv pip install --editable '.[notebooks]'`.\n",
     "See **Run notebooks locally** in the repository README for the kernel, execution order,\n",
-    "and full-data options. Generated artifacts stay under `.cache/notebooks/`.\n"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 1,
-   "id": "f0cae36d",
-   "metadata": {},
-   "outputs": [
-    {
-     "name": "stderr",
-     "output_type": "stream",
-     "text": [
-      "\u001b[38;5;226mEnvironment variable ENVIRONMENT is not set, defaulting to development mode ‼️\u001b[0m\n",
-      "\u001b[38;5;226mThe selected tracing database (ParallelTinyDbDriver) is not recommended for production\u001b[0m\n",
-      "\u001b[38;5;39mGreatAI (v0.1.11): configured ✅\u001b[0m\n",
-      "\u001b[38;5;39m  🔩 tracing_database: ParallelTinyDbDriver\u001b[0m\n",
-      "\u001b[38;5;39m  🔩 large_file_implementation: LargeFileLocal\u001b[0m\n",
-      "\u001b[38;5;39m  🔩 is_production: False\u001b[0m\n",
-      "\u001b[38;5;39m  🔩 should_log_exception_stack: True\u001b[0m\n",
-      "\u001b[38;5;39m  🔩 prediction_cache_size: 512\u001b[0m\n",
-      "\u001b[38;5;39m  🔩 dashboard_table_size: 50\u001b[0m\n"
-     ]
-    },
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "Notebook artifacts: /Users/schmelczer/Projects/great-ai/.cache/notebooks/simple\n"
-     ]
-    }
-   ],
-   "source": [
-    "import os\n",
-    "import sys\n",
-    "from pathlib import Path\n",
-    "\n",
-    "# Works when opened from the repository root or any notebook directory.\n",
-    "_start = Path(globals().get(\"__file__\", Path.cwd())).resolve()\n",
-    "ROOT = next(\n",
-    "    p\n",
-    "    for p in [_start, *_start.parents]\n",
-    "    if (p / \"pyproject.toml\").is_file() and (p / \"great_ai\").is_dir()\n",
-    ")\n",
-    "sys.path.insert(0, str(ROOT / \"docs\"))\n",
-    "from notebook_support import setup_notebook, mag_lines\n",
-    "\n",
-    "FULL_RUN = os.environ.get(\"GREAT_AI_NOTEBOOK_FULL\") == \"1\"\n",
-    "WORK_DIR = setup_notebook(ROOT, \"simple\")\n",
-    "DATA_DIR = ROOT / \"docs/tutorial/data\""
+    "and full-data options. Run related notebooks from the same working directory\n",
+    "to share the default `.cache/` model store and `tracing_database.json` database.\n"
    ]
   },
   {
@@ -126,7 +78,7 @@
     "ax.legend()\n",
     "ax.set_ylabel(\"Ratio of all publications\")\n",
     "fig.tight_layout()\n",
-    "fig.savefig(WORK_DIR / \"ss-distribution.png\", dpi=500)\n",
+    "fig.savefig(\"ss-distribution.png\", dpi=500)\n",
     "None"
    ]
   }
diff --git a/docs/examples/simple/train.ipynb b/docs/examples/simple/train.ipynb
index bdc62ff..1e883cd 100644
--- a/docs/examples/simple/train.ipynb
+++ b/docs/examples/simple/train.ipynb
@@ -26,56 +26,18 @@
     "\n",
     "Use Python 3.12 or 3.13 and install this checkout with `uv pip install --editable '.[notebooks]'`.\n",
     "See **Run notebooks locally** in the repository README for the kernel, execution order,\n",
-    "and full-data options. Generated artifacts stay under `.cache/notebooks/`.\n"
+    "and full-data options. Run related notebooks from the same working directory\n",
+    "to share the default `.cache/` model store and `tracing_database.json` database.\n"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 1,
+   "execution_count": null,
    "id": "d8ae0655",
    "metadata": {},
-   "outputs": [
-    {
-     "name": "stderr",
-     "output_type": "stream",
-     "text": [
-      "\u001b[38;5;226mEnvironment variable ENVIRONMENT is not set, defaulting to development mode ‼️\u001b[0m\n",
-      "\u001b[38;5;226mThe selected tracing database (ParallelTinyDbDriver) is not recommended for production\u001b[0m\n",
-      "\u001b[38;5;39mGreatAI (v0.1.11): configured ✅\u001b[0m\n",
-      "\u001b[38;5;39m  🔩 tracing_database: ParallelTinyDbDriver\u001b[0m\n",
-      "\u001b[38;5;39m  🔩 large_file_implementation: LargeFileLocal\u001b[0m\n",
-      "\u001b[38;5;39m  🔩 is_production: False\u001b[0m\n",
-      "\u001b[38;5;39m  🔩 should_log_exception_stack: True\u001b[0m\n",
-      "\u001b[38;5;39m  🔩 prediction_cache_size: 512\u001b[0m\n",
-      "\u001b[38;5;39m  🔩 dashboard_table_size: 50\u001b[0m\n"
-     ]
-    },
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "Notebook artifacts: /Users/schmelczer/Projects/great-ai/.cache/notebooks/simple\n"
-     ]
-    }
-   ],
+   "outputs": [],
    "source": [
-    "import os\n",
-    "import sys\n",
-    "from pathlib import Path\n",
-    "\n",
-    "# Works when opened from the repository root or any notebook directory.\n",
-    "_start = Path(globals().get(\"__file__\", Path.cwd())).resolve()\n",
-    "ROOT = next(\n",
-    "    p\n",
-    "    for p in [_start, *_start.parents]\n",
-    "    if (p / \"pyproject.toml\").is_file() and (p / \"great_ai\").is_dir()\n",
-    ")\n",
-    "sys.path.insert(0, str(ROOT / \"docs\"))\n",
-    "from notebook_support import setup_notebook, mag_lines\n",
-    "\n",
-    "FULL_RUN = os.environ.get(\"GREAT_AI_NOTEBOOK_FULL\") == \"1\"\n",
-    "WORK_DIR = setup_notebook(ROOT, \"simple\")\n",
-    "DATA_DIR = ROOT / \"docs/tutorial/data\""
+    "from docs.notebook_support import FULL_RUN"
    ]
   },
   {
diff --git a/docs/notebook_support.py b/docs/notebook_support.py
index 1024828..65358f5 100644
--- a/docs/notebook_support.py
+++ b/docs/notebook_support.py
@@ -12,27 +12,6 @@ SURVEY_DIR = ROOT / "docs/surveys"
 FULL_RUN = os.environ.get("GREAT_AI_NOTEBOOK_FULL") == "1"
 
 
-def setup_notebook(workflow: str) -> Path:
-    """Keep related notebooks' models and traces together, outside source files."""
-    from great_ai import ParallelTinyDbDriver, configure
-    from great_ai.large_file import LargeFileLocal
-
-    base = Path(os.environ.get("GREAT_AI_NOTEBOOK_WORKDIR", ROOT / ".cache/notebooks"))
-    suffix = "-full" if FULL_RUN else ""
-    work_dir = (base / (workflow + suffix)).resolve()
-    work_dir.mkdir(parents=True, exist_ok=True)
-    LargeFileLocal.cache_path = work_dir / "models-cache"
-    ParallelTinyDbDriver.path_to_db = work_dir / "tracing_database.json"
-    configure(
-        large_file_implementation=LargeFileLocal,
-        tracing_database_factory=ParallelTinyDbDriver,
-        disable_se4ml_banner=True,
-        seed=42,
-    )
-    print(f"Notebook artifacts: {work_dir}")
-    return work_dir
-
-
 def mag_lines(path: Path, per_label: Optional[int] = None) -> List[str]:
     """Read the bundled MAG data, optionally limiting each class for a quick run."""
     counts: Counter = Counter()
diff --git a/docs/surveys/surveys.ipynb b/docs/surveys/surveys.ipynb
index cb80fe5..f4b7fed 100644
--- a/docs/surveys/surveys.ipynb
+++ b/docs/surveys/surveys.ipynb
@@ -8,7 +8,8 @@
     "\n",
     "Use Python 3.12 or 3.13 and install this checkout with `uv pip install --editable '.[notebooks]'`.\n",
     "See **Run notebooks locally** in the repository README for the kernel, execution order,\n",
-    "and full-data options. Generated artifacts stay under `.cache/notebooks/`.\n"
+    "and full-data options. Run related notebooks from the same working directory\n",
+    "to share the default `.cache/` model store and `tracing_database.json` database.\n"
    ],
    "metadata": {}
   },
@@ -18,23 +19,7 @@
    "metadata": {},
    "execution_count": null,
    "source": [
-    "import os\n",
-    "import sys\n",
-    "from pathlib import Path\n",
-    "\n",
-    "# Works when opened from the repository root or any notebook directory.\n",
-    "_start = Path(globals().get(\"__file__\", Path.cwd())).resolve()\n",
-    "ROOT = next(\n",
-    "    p\n",
-    "    for p in [_start, *_start.parents]\n",
-    "    if (p / \"pyproject.toml\").is_file() and (p / \"great_ai\").is_dir()\n",
-    ")\n",
-    "sys.path.insert(0, str(ROOT / \"docs\"))\n",
-    "from notebook_support import setup_notebook, mag_lines\n",
-    "\n",
-    "FULL_RUN = os.environ.get(\"GREAT_AI_NOTEBOOK_FULL\") == \"1\"\n",
-    "WORK_DIR = setup_notebook(ROOT, \"surveys\")\n",
-    "DATA_DIR = ROOT / \"docs/tutorial/data\""
+    "from docs.notebook_support import SURVEY_DIR"
    ],
    "outputs": []
   },
@@ -46,7 +31,6 @@
    "source": [
     "import pandas as pd\n",
     "\n",
-    "SURVEY_DIR = ROOT / \"docs/surveys\"\n",
     "best_practices = pd.read_csv(SURVEY_DIR / \"Best practices assessment.csv\")\n",
     "best_practices"
    ],
@@ -161,7 +145,7 @@
     ")\n",
     "plt.tight_layout()\n",
     "\n",
-    "plt.savefig(WORK_DIR / \"best-practices.png\", dpi=150)\n",
+    "plt.savefig(\"best-practices.png\", dpi=150)\n",
     "\n",
     "stats.pearsonr(\n",
     "    [x for x, y in best_practice_score_se], [y for x, y in best_practice_score_se]\n",
diff --git a/docs/tutorial/deploy.ipynb b/docs/tutorial/deploy.ipynb
index 33d1767..1937989 100644
--- a/docs/tutorial/deploy.ipynb
+++ b/docs/tutorial/deploy.ipynb
@@ -20,36 +20,11 @@
     "\n",
     "Use Python 3.12 or 3.13 and install this checkout with `uv pip install --editable '.[notebooks]'`.\n",
     "See **Run notebooks locally** in the repository README for the kernel, execution order,\n",
-    "and full-data options. Generated artifacts stay under `.cache/notebooks/`.\n"
+    "and full-data options. Run related notebooks from the same working directory\n",
+    "to share the default `.cache/` model store and `tracing_database.json` database.\n"
    ],
    "metadata": {}
   },
-  {
-   "id": "bf8f9a33",
-   "cell_type": "code",
-   "metadata": {},
-   "execution_count": null,
-   "source": [
-    "import os\n",
-    "import sys\n",
-    "from pathlib import Path\n",
-    "\n",
-    "# Works when opened from the repository root or any notebook directory.\n",
-    "_start = Path(globals().get(\"__file__\", Path.cwd())).resolve()\n",
-    "ROOT = next(\n",
-    "    p\n",
-    "    for p in [_start, *_start.parents]\n",
-    "    if (p / \"pyproject.toml\").is_file() and (p / \"great_ai\").is_dir()\n",
-    ")\n",
-    "sys.path.insert(0, str(ROOT / \"docs\"))\n",
-    "from notebook_support import setup_notebook, mag_lines\n",
-    "\n",
-    "FULL_RUN = os.environ.get(\"GREAT_AI_NOTEBOOK_FULL\") == \"1\"\n",
-    "WORK_DIR = setup_notebook(ROOT, \"tutorial\")\n",
-    "DATA_DIR = ROOT / \"docs/tutorial/data\""
-   ],
-   "outputs": []
-  },
   {
    "cell_type": "code",
    "execution_count": null,
diff --git a/docs/tutorial/train.ipynb b/docs/tutorial/train.ipynb
index b84320d..2c00097 100644
--- a/docs/tutorial/train.ipynb
+++ b/docs/tutorial/train.ipynb
@@ -22,7 +22,8 @@
     "\n",
     "Use Python 3.12 or 3.13 and install this checkout with `uv pip install --editable '.[notebooks]'`.\n",
     "See **Run notebooks locally** in the repository README for the kernel, execution order,\n",
-    "and full-data options. Generated artifacts stay under `.cache/notebooks/`.\n"
+    "and full-data options. Run related notebooks from the same working directory\n",
+    "to share the default `.cache/` model store and `tracing_database.json` database.\n"
    ],
    "metadata": {}
   },
@@ -32,23 +33,7 @@
    "metadata": {},
    "execution_count": null,
    "source": [
-    "import os\n",
-    "import sys\n",
-    "from pathlib import Path\n",
-    "\n",
-    "# Works when opened from the repository root or any notebook directory.\n",
-    "_start = Path(globals().get(\"__file__\", Path.cwd())).resolve()\n",
-    "ROOT = next(\n",
-    "    p\n",
-    "    for p in [_start, *_start.parents]\n",
-    "    if (p / \"pyproject.toml\").is_file() and (p / \"great_ai\").is_dir()\n",
-    ")\n",
-    "sys.path.insert(0, str(ROOT / \"docs\"))\n",
-    "from notebook_support import setup_notebook, mag_lines\n",
-    "\n",
-    "FULL_RUN = os.environ.get(\"GREAT_AI_NOTEBOOK_FULL\") == \"1\"\n",
-    "WORK_DIR = setup_notebook(ROOT, \"tutorial\")\n",
-    "DATA_DIR = ROOT / \"docs/tutorial/data\""
+    "from docs.notebook_support import DATA_DIR, FULL_RUN, mag_lines"
    ],
    "outputs": []
   },
diff --git a/scripts/run-notebooks.py b/scripts/run-notebooks.py
index 337cc51..7fbb6b8 100644
--- a/scripts/run-notebooks.py
+++ b/scripts/run-notebooks.py
@@ -68,10 +68,16 @@ def main() -> None:
     # A fresh workspace detects missing dependencies between notebook stages and
     # avoids silently passing because an old model or database happens to exist.
     with TemporaryDirectory(prefix="great-ai-notebooks-") as workspace:
-        env = dict(os.environ, GREAT_AI_NOTEBOOK_WORKDIR=workspace)
+        env = dict(os.environ)
+        env["PYTHONPATH"] = os.pathsep.join(
+            filter(None, [str(ROOT), env.get("PYTHONPATH")])
+        )
         for name in args.notebooks or NOTEBOOKS:
             path = (ROOT / name).resolve()
             relative = path.relative_to(ROOT)
+            # Related notebooks share the library's default local storage.
+            work_dir = Path(workspace) / relative.parent
+            work_dir.mkdir(parents=True, exist_ok=True)
             print(f"Running {relative}", flush=True)
             notebook = nbformat.read(path, as_version=4)
             nbformat.validate(notebook)
@@ -86,7 +92,7 @@ def main() -> None:
             ]
             client = NotebookClient(notebook, km=manager, timeout=args.timeout)
             try:
-                client.execute(cwd=str(path.parent), env=env, cleanup_kc=True)
+                client.execute(cwd=str(work_dir), env=env, cleanup_kc=True)
             finally:
                 destination = output_dir / relative
                 destination.parent.mkdir(parents=True, exist_ok=True)
@@ -100,7 +106,7 @@ def main() -> None:
                         str(destination),
                         path.parent.name,
                     ],
-                    cwd=ROOT,
+                    cwd=work_dir,
                     env=env,
                     check=True,
                     timeout=args.timeout,
diff --git a/tests/test_notebook_setup.py b/tests/test_notebook_setup.py
deleted file mode 100644
index 2c6af5f..0000000
--- a/tests/test_notebook_setup.py
+++ /dev/null
@@ -1,42 +0,0 @@
-import json
-from pathlib import Path
-from typing import Any, Dict
-
-import pytest
-from great_ai import ParallelTinyDbDriver
-from great_ai.context import get_context
-from great_ai.large_file import LargeFileLocal
-
-ROOT = Path(__file__).resolve().parents[1]
-NOTEBOOKS = sorted((ROOT / "docs").rglob("*.ipynb"))
-
-
-@pytest.mark.parametrize(
-    "notebook", NOTEBOOKS, ids=lambda path: str(path.relative_to(ROOT))
-)
-def test_notebook_setup_uses_checkout_paths_from_another_directory(
-    isolated_context: None,
-    notebook: Path,
-    tmp_path: Path,
-    monkeypatch: pytest.MonkeyPatch,
-) -> None:
-    monkeypatch.syspath_prepend(str(ROOT))
-    from docs import notebook_support
-
-    monkeypatch.chdir(tmp_path)
-    monkeypatch.setenv("GREAT_AI_NOTEBOOK_WORKDIR", str(tmp_path))
-    monkeypatch.setattr(notebook_support, "FULL_RUN", False)
-    monkeypatch.setattr(ParallelTinyDbDriver, "path_to_db", tmp_path / "traces.json")
-    cells = json.loads(notebook.read_text())["cells"]
-    setup = next(
-        "".join(cell["source"]) for cell in cells if cell["cell_type"] == "code"
-    )
-    namespace: Dict[str, Any] = {}
-    exec(compile(setup, str(notebook), "exec"), namespace)
-    work_dir = tmp_path / notebook.parent.name
-    assert work_dir.is_dir()
-    assert LargeFileLocal.cache_path == work_dir / "models-cache"
-    database = get_context().tracing_database
-    assert isinstance(database, ParallelTinyDbDriver)
-    assert database.path_to_db == work_dir / "tracing_database.json"
-    assert (notebook_support.DATA_DIR / "train.txt").is_file()