diff --git a/docs/poli-docs/using_poli/the_basics/defining_an_observer.ipynb b/docs/poli-docs/using_poli/the_basics/defining_an_observer.ipynb
index 49db7fd1..a331c60d 100644
--- a/docs/poli-docs/using_poli/the_basics/defining_an_observer.ipynb
+++ b/docs/poli-docs/using_poli/the_basics/defining_an_observer.ipynb
@@ -34,7 +34,7 @@
    "metadata": {},
    "source": [
     "All observers inherit from an `AbstractObserver` (which you can find on `poli/core/util/abstract_observer.py`). The abstract methods you need to overwrite are:\n",
-    "- `initialize_observer(problem_info: ProblemSetupInformation, caller_info: object, x0: np.ndarray, y0: np.ndarray, seed: int) -> object`, which gets called as part of the set-up of the objective function (when `objective_factory.create` is called).\n",
+    "- `initialize_observer(problem_info: BlackBoxInformation, caller_info: object, seed: int) -> object`, which gets called as part of the set-up of the objective function (when `objective_factory.create` is called).\n",
     "- `observe(x: np.ndarray, y: np.ndarray, context: object) -> None`, which gets called every time your optimization algorithms query the objective function.\n",
     "- `finish()`, which gets called either by the user, or by the object deletion at the end of the script."
    ]
@@ -61,7 +61,7 @@
    "source": [
     "import numpy as np\n",
     "\n",
-    "from poli.core.problem_setup_information import ProblemSetupInformation\n",
+    "from poli.core.black_box_information import BlackBoxInformation\n",
     "from poli.core.util.abstract_observer import AbstractObserver\n",
     "\n",
     "class SimpleObserver(AbstractObserver):\n",
@@ -70,10 +70,8 @@
     "\n",
     "    def initialize_observer(\n",
     "        self,\n",
-    "        problem_setup_info: ProblemSetupInformation,\n",
+    "        problem_setup_info: BlackBoxInformation,\n",
     "        caller_info: object,\n",
-    "        x0: np.ndarray,\n",
-    "        y0: np.ndarray,\n",
     "        seed: int\n",
     "    ) -> object:\n",
     "        ...\n",
@@ -111,7 +109,7 @@
     "\n",
     "import numpy as np\n",
     "\n",
-    "from poli.core.problem_setup_information import ProblemSetupInformation\n",
+    "from poli.core.black_box_information import BlackBoxInformation\n",
     "from poli.core.util.abstract_observer import AbstractObserver\n",
     "\n",
     "THIS_DIR = Path().resolve()\n",
@@ -131,10 +129,8 @@
     "    \n",
     "    def initialize_observer(\n",
     "        self,\n",
-    "        problem_setup_info: ProblemSetupInformation,\n",
+    "        problem_setup_info: BlackBoxInformation,\n",
     "        caller_info: object,\n",
-    "        x0: np.ndarray,\n",
-    "        y0: np.ndarray,\n",
     "        seed: int\n",
     "    ) -> object:\n",
     "\n",
@@ -145,10 +141,6 @@
     "        # (Recall that this caller info gets forwarded\n",
     "        # from the objective_factory.create function)\n",
     "        metadata[\"caller_info\"] = caller_info\n",
-    "\n",
-    "        # Saving the initial evaluations and seed\n",
-    "        metadata[\"x0\"] = x0.tolist()\n",
-    "        metadata[\"y0\"] = y0.tolist()\n",
     "        metadata[\"seed\"] = seed\n",
     "\n",
     "        # Saving the metadata\n",
@@ -193,7 +185,6 @@
     "\n",
     "import numpy as np\n",
     "\n",
-    "from poli.core.problem_setup_information import ProblemSetupInformation\n",
     "from poli.core.util.abstract_observer import AbstractObserver\n",
     "\n",
     "THIS_DIR = Path().resolve()\n",
@@ -238,7 +229,7 @@
     "\n",
     "import numpy as np\n",
     "\n",
-    "from poli.core.problem_setup_information import ProblemSetupInformation\n",
+    "from poli.core.black_box_information import BlackBoxInformation\n",
     "from poli.core.util.abstract_observer import AbstractObserver\n",
     "\n",
     "THIS_DIR = Path().resolve()\n",
@@ -258,10 +249,8 @@
     "    \n",
     "    def initialize_observer(\n",
     "        self,\n",
-    "        problem_setup_info: ProblemSetupInformation,\n",
+    "        problem_setup_info: BlackBoxInformation,\n",
     "        caller_info: object,\n",
-    "        x0: np.ndarray,\n",
-    "        y0: np.ndarray,\n",
     "        seed: int\n",
     "    ) -> object:\n",
     "\n",
@@ -288,6 +277,32 @@
     "            fp.write(f\"{x.tolist()}\\t{y.tolist()}\\n\")"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Registration\n",
+    "To profit from environment isolation, users have to register their observers (**once**)."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 5,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from poli.core.registry import register_observer\n",
+    "\n",
+    "from simple_observer import SimpleObserver\n",
+    "\n",
+    "register_observer(\n",
+    "    observer=SimpleObserver(),\n",
+    "    # conda_environment_location=\"poli\",  # when not providing the environment, we use the current one\n",
+    "    observer_name=\"simple_observer\",\n",
+    "    set_as_default_observer=False,  # this is True by default!\n",
+    ")"
+   ]
+  },
   {
    "cell_type": "markdown",
    "metadata": {},
@@ -304,12 +319,8 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 5,
-   "metadata": {
-    "tags": [
-     "hide-output"
-    ]
-   },
+   "execution_count": 6,
+   "metadata": {},
    "outputs": [
     {
      "name": "stdout",
@@ -323,13 +334,10 @@
    "source": [
     "from poli import objective_factory\n",
     "\n",
-    "# We create an instance of the observer\n",
-    "observer = SimpleObserver()\n",
-    "\n",
     "# We instantiate the objective function\n",
     "problem = objective_factory.create(\n",
     "    name=\"aloha\",\n",
-    "    observer=observer,\n",
+    "    observer_name=\"simple_observer\",  # instantiate the registered observer\n",
     ")\n",
     "f, x0 = problem.black_box, problem.x0"
    ]
@@ -338,82 +346,104 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "At this point, the observer `__init__` call created a folder called `results` right next to this file, and we can load up the metadata just to be sure:"
+    "Let's query the objective function at three points, and check whether the results were saved accordingly: "
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 6,
+   "execution_count": 8,
    "metadata": {},
    "outputs": [
     {
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "{'name': 'aloha', 'max_sequence_length': 5, 'aligned': True, 'fixed_length': True, 'deterministic': True, 'discrete': True, 'fidelity': None, 'alphabet': ['A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', 'I', 'J', 'K', 'L', 'M', 'N', 'O', 'P', 'Q', 'R', 'S', 'T', 'U', 'V', 'W', 'X', 'Y', 'Z'], 'log_transform_recommended': False, 'padding_token': '', 'caller_info': None, 'x0': [['A', 'L', 'O', 'O', 'F']], 'y0': [[3]], 'seed': None}\n"
+      "[[0]]\n",
+      "[[1]]\n",
+      "[[5]]\n"
      ]
     }
    ],
    "source": [
-    "with open(observer.experiment_path / \"metadata.json\") as fp:\n",
-    "    print(json.load(fp))"
+    "print(f(np.array([list(\"MIGUE\")])))\n",
+    "print(f(np.array([list(\"FLEAS\")])))\n",
+    "print(f(np.array([list(\"ALOHA\")])))"
    ]
   },
   {
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "Let's query the objective function at three points, and check whether the results were saved accordingly: "
+    "We can verify by loading up and printing the `results.txt` file:"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 7,
+   "execution_count": 9,
    "metadata": {},
    "outputs": [
     {
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "[[0]]\n",
-      "[[1]]\n",
-      "[[5]]\n"
+      "[['M', 'I', 'G', 'U', 'E']]\t[[0]]\n",
+      "[['F', 'L', 'E', 'A', 'S']]\t[[1]]\n",
+      "[['A', 'L', 'O', 'H', 'A']]\t[[5]]\n",
+      "\n"
      ]
     }
    ],
    "source": [
-    "print(f(np.array([list(\"MIGUE\")])))\n",
-    "print(f(np.array([list(\"FLEAS\")])))\n",
-    "print(f(np.array([list(\"ALOHA\")])))"
+    "with open(f.observer.experiment_path / \"results.txt\") as fp:\n",
+    "    print(fp.read())"
    ]
   },
   {
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "We can verify by loading up and printing the `results.txt` file:"
+    "### Discouraged Use\n",
+    "The example below shows what happens behind the curtains in `poli` (after dynamic instantiation). Some users may "
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 8,
-   "metadata": {},
+   "execution_count": 5,
+   "metadata": {
+    "scrolled": true,
+    "tags": [
+     "hide-output"
+    ]
+   },
    "outputs": [
     {
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "[['A', 'L', 'O', 'O', 'F']]\t[[3]]\n",
-      "[['M', 'I', 'G', 'U', 'E']]\t[[0]]\n",
-      "[['F', 'L', 'E', 'A', 'S']]\t[[1]]\n",
-      "[['A', 'L', 'O', 'H', 'A']]\t[[5]]\n",
-      "\n"
+      "poli 🧪: Creating the objective aloha from the repository.\n",
+      "poli 🧪: initializing the observer.\n"
      ]
     }
    ],
    "source": [
-    "with open(observer.experiment_path / \"results.txt\") as fp:\n",
-    "    print(fp.read())"
+    "from poli import objective_factory\n",
+    "from poli.core.registry import DEFAULT_OBSERVER_NAME\n",
+    "\n",
+    "# We create an instance of the observer\n",
+    "observer = SimpleObserver()\n",
+    "\n",
+    "# We instantiate the objective function\n",
+    "problem = objective_factory.create(\n",
+    "    name=\"aloha\",\n",
+    "    observer_name=DEFAULT_OBSERVER_NAME,  # instantiates the default observer which does nothing\n",
+    ")\n",
+    "f, x0 = problem.black_box, problem.x0\n",
+    "\n",
+    "# Then we initialize our observer...\n",
+    "observer.initialize_observer(seed=0, problem_setup_info=problem.black_box_information, caller_info=dict())\n",
+    "\n",
+    "#... and register it with the blackbox.\n",
+    "f.set_observer(observer)"
    ]
   },
   {
@@ -436,16 +466,13 @@
     ":::{tip}\n",
     "\n",
     "If you are interested in using more complex logic for your logging, you can check the `examples` folder in `poli`, as they include two observers using `mlflow` and `wandb`.\n",
-    "\n",
-    "`poli` is also able to isolate observers, and the `examples` folder also includes a description of how.\n",
-    "\n",
     ":::\n"
    ]
   }
  ],
  "metadata": {
   "kernelspec": {
-   "display_name": "poli-base",
+   "display_name": "Python 3 (ipykernel)",
    "language": "python",
    "name": "python3"
   },
@@ -459,10 +486,9 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.9.18"
-  },
-  "orig_nbformat": 4
+   "version": "3.9.19"
+  }
  },
  "nbformat": 4,
- "nbformat_minor": 2
+ "nbformat_minor": 4
 }
diff --git a/docs/poli-docs/using_poli/the_basics/simple_observer.py b/docs/poli-docs/using_poli/the_basics/simple_observer.py
new file mode 100644
index 00000000..4a058699
--- /dev/null
+++ b/docs/poli-docs/using_poli/the_basics/simple_observer.py
@@ -0,0 +1,49 @@
+from pathlib import Path
+from uuid import uuid4
+import json
+
+import numpy as np
+
+from poli.core.black_box_information import BlackBoxInformation
+from poli.core.util.abstract_observer import AbstractObserver
+
+THIS_DIR = Path().resolve()
+
+class SimpleObserver(AbstractObserver):
+    def __init__(self):
+        # Creating a unique id for this experiment in
+        # particular:
+        experiment_id = str(uuid4())
+        self.experiment_id = experiment_id
+
+        # Creating a local directory for the results
+        experiment_path = THIS_DIR / "results" / experiment_id
+        experiment_path.mkdir(exist_ok=True, parents=True)
+        
+        self.experiment_path = experiment_path
+    
+    def initialize_observer(
+        self,
+        problem_setup_info: BlackBoxInformation,
+        caller_info: object,
+        seed: int
+    ) -> object:
+
+        # Saving the metadata for this experiment
+        metadata = problem_setup_info.as_dict()
+
+        # Adding the information the user wanted to provide
+        # (Recall that this caller info gets forwarded
+        # from the objective_factory.create function)
+        metadata["caller_info"] = caller_info
+
+        metadata["seed"] = seed
+
+        # Saving the metadata
+        with open(self.experiment_path / "metadata.json", "w") as f:
+            json.dump(metadata, f)
+    
+    def observe(self, x: np.ndarray, y: np.ndarray, context=None) -> None:
+        # Appending these results to the results file.
+        with open(self.experiment_path / "results.txt", "a") as fp:
+            fp.write(f"{x.tolist()}\t{y.tolist()}\n")