diff --git a/README.md b/README.md
index d393dccb..7dd991bd 100644
--- a/README.md
+++ b/README.md
@@ -137,6 +137,8 @@ import pandas as pd
 from timecopilot import TimeCopilot
 
 # Load the dataset
+# TimeCopilot accepts pandas DataFrames or Polars DataFrames (via narwhals); inputs
+# are normalized to pandas before modeling.
 # The DataFrame must include at least the following columns:
 # - unique_id: Unique identifier for each time series (string)
 # - ds: Date column (datetime format)
diff --git a/docs/getting-started/quickstart.md b/docs/getting-started/quickstart.md
index 3df9eefa..4c6825db 100644
--- a/docs/getting-started/quickstart.md
+++ b/docs/getting-started/quickstart.md
@@ -78,6 +78,8 @@ import pandas as pd
 from timecopilot import TimeCopilot
 
 # Load the dataset
+# TimeCopilot accepts pandas DataFrames or Polars DataFrames (via narwhals); inputs
+# are normalized to pandas before modeling.
 # The DataFrame must include at least the following columns:
 # - unique_id: Unique identifier for each time series (string)
 # - ds: Date column (datetime format)
diff --git a/pyproject.toml b/pyproject.toml
index e995d9db..a032353a 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -54,6 +54,7 @@ dependencies = [
   "lightgbm>=4.6.0",
   "logfire>=4.7.0",
   "mlforecast>=1.0.2",
+  "narwhals>=2.7.0",
   "neuralforecast>=3.0.2",
   "nixtla>=0.6.6",
   "openai>=1.99.7",
diff --git a/tests/utils/test_df_utils.py b/tests/utils/test_df_utils.py
new file mode 100644
index 00000000..e605957a
--- /dev/null
+++ b/tests/utils/test_df_utils.py
@@ -0,0 +1,87 @@
+from importlib import util
+from pathlib import Path
+
+import narwhals as nw
+import pandas as pd
+import pytest
+
+
+def load_to_pandas():
+    module_path = (
+        Path(__file__).resolve().parents[2] / "timecopilot" / "utils" / "df_utils.py"
+    )
+    spec = util.spec_from_file_location("df_utils", module_path)
+    if spec is None or spec.loader is None:
+        raise RuntimeError("Failed to load df_utils module.")
+    module = util.module_from_spec(spec)
+    spec.loader.exec_module(module)
+    return module.to_pandas
+
+
+def load_experiment_handler():
+    module_path = (
+        Path(__file__).resolve().parents[2]
+        / "timecopilot"
+        / "utils"
+        / "experiment_handler.py"
+    )
+    spec = util.spec_from_file_location("experiment_handler", module_path)
+    if spec is None or spec.loader is None:
+        raise RuntimeError("Failed to load experiment_handler module.")
+    module = util.module_from_spec(spec)
+    spec.loader.exec_module(module)
+    return module
+
+
+def test_to_pandas_accepts_pandas_dataframe():
+    df = pd.DataFrame({"unique_id": ["a"], "ds": ["2024-01-01"], "y": [1.0]})
+    to_pandas = load_to_pandas()
+    result = to_pandas(df)
+    assert isinstance(result, pd.DataFrame)
+    pd.testing.assert_frame_equal(result, df)
+
+
+def test_to_pandas_accepts_narwhals_dataframe():
+    nw_df = nw.from_dict(
+        {"unique_id": ["a"], "ds": ["2024-01-01"], "y": [1.0]},
+        backend="pandas",
+    )
+    to_pandas = load_to_pandas()
+    result = to_pandas(nw_df)
+    assert isinstance(result, pd.DataFrame)
+    assert list(result.columns) == ["unique_id", "ds", "y"]
+
+
+def test_to_pandas_accepts_polars_dataframe():
+    pl = pytest.importorskip("polars")
+    pytest.importorskip("pyarrow")
+
+    pl_df = pl.DataFrame({"unique_id": ["a"], "ds": ["2024-01-01"], "y": [1.0]})
+    to_pandas = load_to_pandas()
+    result = to_pandas(pl_df)
+    assert isinstance(result, pd.DataFrame)
+    assert list(result.columns) == ["unique_id", "ds", "y"]
+
+
+def test_to_pandas_fallback_for_unknown_dataframe():
+    fallback = pd.DataFrame({"unique_id": ["fallback"], "ds": ["2024-01-01"], "y": [0.0]})
+    to_pandas = load_to_pandas()
+    result = to_pandas(object(), fallback=fallback)
+    pd.testing.assert_frame_equal(result, fallback)
+
+
+def test_to_pandas_raises_for_unknown_dataframe():
+    to_pandas = load_to_pandas()
+    with pytest.raises(TypeError, match="Unsupported dataframe type"):
+        to_pandas(object())
+
+
+def test_validate_df_accepts_polars_dataframe():
+    pl = pytest.importorskip("polars")
+    pytest.importorskip("pyarrow")
+
+    df = pl.DataFrame({"unique_id": ["a"], "ds": ["2024-01-01"], "y": [1.0]})
+
+    module = load_experiment_handler()
+    parsed = module.ExperimentDatasetParser._validate_df(df)
+    assert isinstance(parsed, pd.DataFrame)
diff --git a/timecopilot/forecaster.py b/timecopilot/forecaster.py
index a27ac5f4..0184d02b 100644
--- a/timecopilot/forecaster.py
+++ b/timecopilot/forecaster.py
@@ -1,6 +1,7 @@
 import pandas as pd
 
 from .models.utils.forecaster import Forecaster
+from .utils.df_utils import to_pandas
 
 
 class TimeCopilotForecaster(Forecaster):
@@ -73,6 +74,7 @@ def _call_models(
         quantiles: list[float] | None,
         **kwargs,
     ) -> pd.DataFrame:
+        df = to_pandas(df)
         # infer just once to avoid multiple calls to _maybe_infer_freq
         freq = self._maybe_infer_freq(df, freq)
         res_df: pd.DataFrame | None = None
@@ -173,6 +175,7 @@ def forecast(
                 For multi-series data, the output retains the same unique
                 identifiers as the input DataFrame.
         """
+        df = to_pandas(df)
         return self._call_models(
             "forecast",
             merge_on=["unique_id", "ds"],
@@ -247,6 +250,7 @@ def cross_validation(
                     - prediction intervals if `level` is specified.
                     - quantile forecasts if `quantiles` is specified.
         """
+        df = to_pandas(df)
         return self._call_models(
             "cross_validation",
             merge_on=["unique_id", "ds", "cutoff"],
@@ -316,6 +320,7 @@ def detect_anomalies(
                         an anomaly is defined as a value that is outside of the
                         prediction interval (True or False).
         """
+        df = to_pandas(df)
         return self._call_models(
             "detect_anomalies",
             merge_on=["unique_id", "ds", "cutoff"],
diff --git a/timecopilot/models/adapters/sktime.py b/timecopilot/models/adapters/sktime.py
index a258226a..dde500af 100644
--- a/timecopilot/models/adapters/sktime.py
+++ b/timecopilot/models/adapters/sktime.py
@@ -4,6 +4,7 @@
 import pandas as pd
 
 from ..utils.forecaster import Forecaster
+from ..utils.df_utils import to_pandas
 
 # from sktime.forecasting.base import BaseForecaster, ForecastingHorizon
 
@@ -128,6 +129,7 @@ def forecast(
                 "Level and quantiles are not supported for adapted sktime models yet."
             )
         # NOTE: may not be needed
+        df = to_pandas(df)
         freq = self._maybe_infer_freq(df, freq)
         forecast_horizon = np.arange(1, 1 + h)
         id_col = "unique_id"
diff --git a/timecopilot/models/ensembles/median.py b/timecopilot/models/ensembles/median.py
index 7b9335e8..ff69bd99 100644
--- a/timecopilot/models/ensembles/median.py
+++ b/timecopilot/models/ensembles/median.py
@@ -3,6 +3,7 @@
 
 from ... import TimeCopilotForecaster
 from ..utils.forecaster import Forecaster, QuantileConverter
+from ..utils.df_utils import to_pandas
 
 
 class MedianEnsemble(Forecaster):
@@ -116,6 +117,7 @@ def forecast(
                 For multi-series data, the output retains the same unique
                 identifiers as the input DataFrame.
         """
+        df = to_pandas(df)
         qc = QuantileConverter(level=level, quantiles=quantiles)
         _fcst_df = self.tcf._call_models(
             "forecast",
diff --git a/timecopilot/models/foundation/chronos.py b/timecopilot/models/foundation/chronos.py
index 95b29062..98e4785f 100644
--- a/timecopilot/models/foundation/chronos.py
+++ b/timecopilot/models/foundation/chronos.py
@@ -12,6 +12,7 @@
 from tqdm import tqdm
 
 from ..utils.forecaster import Forecaster, QuantileConverter
+from ...utils.df_utils import to_pandas
 from .utils import TimeSeriesDataset
 
 
@@ -216,6 +217,7 @@ def forecast(
                 For multi-series data, the output retains the same unique
                 identifiers as the input DataFrame.
         """
+        df = to_pandas(df)
         freq = self._maybe_infer_freq(df, freq)
         qc = QuantileConverter(level=level, quantiles=quantiles)
         dataset = TimeSeriesDataset.from_df(df, batch_size=self.batch_size)
diff --git a/timecopilot/models/foundation/flowstate.py b/timecopilot/models/foundation/flowstate.py
index de9d95b8..5587e33a 100644
--- a/timecopilot/models/foundation/flowstate.py
+++ b/timecopilot/models/foundation/flowstate.py
@@ -9,6 +9,7 @@
 from tsfm_public.models.flowstate.utils.utils import get_fixed_factor
 
 from ..utils.forecaster import Forecaster, QuantileConverter
+from ...utils.df_utils import to_pandas
 from .utils import TimeSeriesDataset
 
 
@@ -257,6 +258,7 @@ def forecast(
                 For multi-series data, the output retains the same unique
                 identifiers as the input DataFrame.
         """
+        df = to_pandas(df)
         freq = self._maybe_infer_freq(df, freq)
         qc = QuantileConverter(level=level, quantiles=quantiles)
         dataset = TimeSeriesDataset.from_df(
diff --git a/timecopilot/models/foundation/sundial.py b/timecopilot/models/foundation/sundial.py
index 3ee6a727..0e32f201 100644
--- a/timecopilot/models/foundation/sundial.py
+++ b/timecopilot/models/foundation/sundial.py
@@ -12,6 +12,7 @@
 from transformers import AutoModelForCausalLM
 
 from ..utils.forecaster import Forecaster, QuantileConverter
+from ...utils.df_utils import to_pandas
 from .utils import TimeSeriesDataset
 
 
@@ -257,6 +258,7 @@ def forecast(
                 For multi-series data, the output retains the same unique
                 identifiers as the input DataFrame.
         """
+        df = to_pandas(df)
         freq = self._maybe_infer_freq(df, freq)
         qc = QuantileConverter(level=level, quantiles=quantiles)
         dataset = TimeSeriesDataset.from_df(df, batch_size=self.batch_size)
diff --git a/timecopilot/models/foundation/tabpfn.py b/timecopilot/models/foundation/tabpfn.py
index 44b5cd02..8cbb53b1 100644
--- a/timecopilot/models/foundation/tabpfn.py
+++ b/timecopilot/models/foundation/tabpfn.py
@@ -26,6 +26,7 @@
 )
 
 from ..utils.forecaster import Forecaster, QuantileConverter
+from ...utils.df_utils import to_pandas
 
 
 class TabPFN(Forecaster):
@@ -199,6 +200,7 @@ def forecast(
                 For multi-series data, the output retains the same unique
                 identifiers as the input DataFrame.
         """
+        df = to_pandas(df)
         freq = self._maybe_infer_freq(df, freq)
         qc = QuantileConverter(level=level, quantiles=quantiles)
         if qc.quantiles is not None and not np.allclose(
diff --git a/timecopilot/models/foundation/timegpt.py b/timecopilot/models/foundation/timegpt.py
index ac888105..2ebbc6a4 100644
--- a/timecopilot/models/foundation/timegpt.py
+++ b/timecopilot/models/foundation/timegpt.py
@@ -4,6 +4,7 @@
 from nixtla import NixtlaClient
 
 from ..utils.forecaster import Forecaster
+from ...utils.df_utils import to_pandas
 
 
 class TimeGPT(Forecaster):
@@ -129,6 +130,7 @@ def forecast(
                 For multi-series data, the output retains the same unique
                 identifiers as the input DataFrame.
         """
+        df = to_pandas(df)
         freq = self._maybe_infer_freq(df, freq)
         client = self._get_client()
         fcst_df = client.forecast(
diff --git a/timecopilot/models/foundation/timesfm.py b/timecopilot/models/foundation/timesfm.py
index 66e89e00..c3b295aa 100644
--- a/timecopilot/models/foundation/timesfm.py
+++ b/timecopilot/models/foundation/timesfm.py
@@ -12,6 +12,7 @@
 from tqdm import tqdm
 
 from ..utils.forecaster import Forecaster, QuantileConverter
+from ...utils.df_utils import to_pandas
 from .utils import TimeSeriesDataset
 
 
@@ -88,6 +89,7 @@ def forecast(
         level: list[int | float] | None = None,
         quantiles: list[float] | None = None,
     ) -> pd.DataFrame:
+        df = to_pandas(df)
         freq = self._maybe_infer_freq(df, freq)
         qc = QuantileConverter(level=level, quantiles=quantiles)
         if qc.quantiles is not None and len(qc.quantiles) != len(DEFAULT_QUANTILES_TFM):
@@ -197,6 +199,7 @@ def forecast(
         level: list[int | float] | None = None,
         quantiles: list[float] | None = None,
     ) -> pd.DataFrame:
+        df = to_pandas(df)
         freq = self._maybe_infer_freq(df, freq)
         qc = QuantileConverter(level=level, quantiles=quantiles)
         if qc.quantiles is not None and len(qc.quantiles) != len(DEFAULT_QUANTILES_TFM):
diff --git a/timecopilot/models/foundation/tirex.py b/timecopilot/models/foundation/tirex.py
index e370c70f..501dc81d 100644
--- a/timecopilot/models/foundation/tirex.py
+++ b/timecopilot/models/foundation/tirex.py
@@ -13,6 +13,7 @@
 from tqdm import tqdm
 
 from ..utils.forecaster import Forecaster, QuantileConverter
+from ...utils.df_utils import to_pandas
 from .utils import TimeSeriesDataset
 
 
@@ -167,6 +168,7 @@ def forecast(
                 For multi-series data, the output retains the same unique
                 identifiers as the input DataFrame.
         """
+        df = to_pandas(df)
         freq = self._maybe_infer_freq(df, freq)
         qc = QuantileConverter(level=level, quantiles=quantiles)
         dataset = TimeSeriesDataset.from_df(df, batch_size=self.batch_size)
diff --git a/timecopilot/models/foundation/toto.py b/timecopilot/models/foundation/toto.py
index 735c4c59..e6184bd7 100644
--- a/timecopilot/models/foundation/toto.py
+++ b/timecopilot/models/foundation/toto.py
@@ -9,6 +9,7 @@
 from tqdm import tqdm
 
 from ..utils.forecaster import Forecaster, QuantileConverter
+from ...utils.df_utils import to_pandas
 from .utils import TimeSeriesDataset
 
 
@@ -222,6 +223,7 @@ def forecast(
                 For multi-series data, the output retains the same unique
                 identifiers as the input DataFrame.
         """
+        df = to_pandas(df)
         freq = self._maybe_infer_freq(df, freq)
         qc = QuantileConverter(level=level, quantiles=quantiles)
         dataset = TimeSeriesDataset.from_df(df, batch_size=self.batch_size)
diff --git a/timecopilot/models/foundation/utils.py b/timecopilot/models/foundation/utils.py
index 287be321..33a76364 100644
--- a/timecopilot/models/foundation/utils.py
+++ b/timecopilot/models/foundation/utils.py
@@ -4,6 +4,8 @@
 import torch
 from utilsforecast.processing import make_future_dataframe
 
+from ...utils.df_utils import to_pandas
+
 
 class TimeSeriesDataset:
     def __init__(
@@ -29,6 +31,7 @@ def from_df(
         batch_size: int,
         dtype: torch.dtype = torch.bfloat16,
     ):
+        df = to_pandas(df)
         tensors = []
         df_sorted = df.sort_values(by=["unique_id", "ds"])
         for _, group in df_sorted.groupby("unique_id"):
diff --git a/timecopilot/models/ml.py b/timecopilot/models/ml.py
index e247f5c8..f7144f64 100644
--- a/timecopilot/models/ml.py
+++ b/timecopilot/models/ml.py
@@ -4,6 +4,7 @@
 from mlforecast.auto import AutoLightGBM, AutoMLForecast
 
 from .utils.forecaster import Forecaster, get_seasonality
+from ..utils.df_utils import to_pandas
 
 os.environ["NIXTLA_ID_AS_COL"] = "true"
 
@@ -84,6 +85,7 @@ def forecast(
         if level is not None or quantiles is not None:
             raise ValueError("Level and quantiles are not supported for AutoLGBM yet.")
 
+        df = to_pandas(df)
         freq = self._maybe_infer_freq(df, freq)
         mf = AutoMLForecast(
             models=[AutoLightGBM()],
diff --git a/timecopilot/models/neural.py b/timecopilot/models/neural.py
index dcc24395..92e6cb07 100644
--- a/timecopilot/models/neural.py
+++ b/timecopilot/models/neural.py
@@ -12,6 +12,7 @@
 from ray import tune
 
 from .utils.forecaster import Forecaster
+from ..utils.df_utils import to_pandas
 
 os.environ["NIXTLA_ID_AS_COL"] = "true"
 
@@ -108,6 +109,7 @@ def forecast(
         if level is not None or quantiles is not None:
             raise ValueError("Level and quantiles are not supported for AutoNHITS yet.")
 
+        df = to_pandas(df)
         inferred_freq = self._maybe_infer_freq(df, freq)
         if self.config is None:
             config = _AutoNHITS.get_default_config(h=h, backend="ray")
@@ -209,6 +211,7 @@ def forecast(
         if level is not None or quantiles is not None:
             raise ValueError("Level and quantiles are not supported for AutoTFT yet.")
 
+        df = to_pandas(df)
         inferred_freq = self._maybe_infer_freq(df, freq)
         if self.config is None:
             config = _AutoTFT.get_default_config(h=h, backend="ray")
diff --git a/timecopilot/models/stats.py b/timecopilot/models/stats.py
index e5a7b742..416c13d4 100644
--- a/timecopilot/models/stats.py
+++ b/timecopilot/models/stats.py
@@ -41,6 +41,7 @@
 from statsforecast.utils import ConformalIntervals
 
 from .utils.forecaster import Forecaster, QuantileConverter, get_seasonality
+from ..utils.df_utils import to_pandas
 
 os.environ["NIXTLA_ID_AS_COL"] = "true"
 
@@ -144,6 +145,7 @@ def forecast(
                 For multi-series data, the output retains the same unique
                 identifiers as the input DataFrame.
         """
+        df = to_pandas(df)
         inferred_freq = self._maybe_infer_freq(df, freq)
         fcst_df = run_statsforecast_model(
             model=_ADIDA(alias=self.alias),
@@ -329,6 +331,7 @@ def forecast(
                 For multi-series data, the output retains the same unique
                 identifiers as the input DataFrame.
         """
+        df = to_pandas(df)
         inferred_freq = self._maybe_infer_freq(df, freq)
         season_length = self._maybe_get_seasonality(inferred_freq)
         fcst_df = run_statsforecast_model(
@@ -458,6 +461,7 @@ def forecast(
                 For multi-series data, the output retains the same unique
                 identifiers as the input DataFrame.
         """
+        df = to_pandas(df)
         inferred_freq = self._maybe_infer_freq(df, freq)
         season_length = self._maybe_get_seasonality(inferred_freq)
         fcst_df = run_statsforecast_model(
@@ -564,6 +568,7 @@ def forecast(
                 For multi-series data, the output retains the same unique
                 identifiers as the input DataFrame.
         """
+        df = to_pandas(df)
         inferred_freq = self._maybe_infer_freq(df, freq)
         season_length = self._maybe_get_seasonality(inferred_freq)
         fcst_df = run_statsforecast_model(
@@ -656,6 +661,7 @@ def forecast(
                 For multi-series data, the output retains the same unique
                 identifiers as the input DataFrame.
         """
+        df = to_pandas(df)
         inferred_freq = self._maybe_infer_freq(df, freq)
         fcst_df = run_statsforecast_model(
             model=_CrostonClassic(
@@ -744,6 +750,7 @@ def forecast(
                 For multi-series data, the output retains the same unique
                 identifiers as the input DataFrame.
         """
+        df = to_pandas(df)
         inferred_freq = self._maybe_infer_freq(df, freq)
         season_length = self._maybe_get_seasonality(inferred_freq)
         fcst_df = run_statsforecast_model(
@@ -833,6 +840,7 @@ def forecast(
                 For multi-series data, the output retains the same unique
                 identifiers as the input DataFrame.
         """
+        df = to_pandas(df)
         inferred_freq = self._maybe_infer_freq(df, freq)
         fcst_df = run_statsforecast_model(
             model=_HistoricAverage(
@@ -921,6 +929,7 @@ def forecast(
                 For multi-series data, the output retains the same unique
                 identifiers as the input DataFrame.
         """
+        df = to_pandas(df)
         inferred_freq = self._maybe_infer_freq(df, freq)
         fcst_df = run_statsforecast_model(
             model=_IMAPA(
@@ -1009,6 +1018,7 @@ def forecast(
                 For multi-series data, the output retains the same unique
                 identifiers as the input DataFrame.
         """
+        df = to_pandas(df)
         inferred_freq = self._maybe_infer_freq(df, freq)
         season_length = self._maybe_get_seasonality(inferred_freq)
         fcst_df = run_statsforecast_model(
@@ -1099,6 +1109,7 @@ def forecast(
                 For multi-series data, the output retains the same unique
                 identifiers as the input DataFrame.
         """
+        df = to_pandas(df)
         inferred_freq = self._maybe_infer_freq(df, freq)
         season_length = self._maybe_get_seasonality(inferred_freq)
         fcst_df = run_statsforecast_model(
@@ -1188,6 +1199,7 @@ def forecast(
                 For multi-series data, the output retains the same unique
                 identifiers as the input DataFrame.
         """
+        df = to_pandas(df)
         inferred_freq = self._maybe_infer_freq(df, freq)
         fcst_df = run_statsforecast_model(
             model=_ZeroModel(
diff --git a/timecopilot/models/utils/forecaster.py b/timecopilot/models/utils/forecaster.py
index 42322bbc..a666d114 100644
--- a/timecopilot/models/utils/forecaster.py
+++ b/timecopilot/models/utils/forecaster.py
@@ -23,6 +23,8 @@
 )
 from utilsforecast.validation import ensure_time_dtype
 
+from ...utils.df_utils import to_pandas
+
 
 def get_seasonality(
     freq: str,
@@ -74,6 +76,7 @@ def maybe_infer_freq(df: pd.DataFrame, freq: str | None) -> str:
     # based on https://github.com/Nixtla/nixtla/blob/bf67c76fd473a61c72b1f54725ffbcb51a3048c5/nixtla/nixtla_client.py#L208C1-L235C25
     if freq is not None:
         return freq
+    df = to_pandas(df)
     sizes = df["unique_id"].value_counts(sort=True)
     times = df.loc[df["unique_id"] == sizes.index[0], "ds"].sort_values()
     if times.dt.tz is not None:
@@ -89,6 +92,7 @@ def maybe_infer_freq(df: pd.DataFrame, freq: str | None) -> str:
 
 
 def maybe_convert_col_to_datetime(df: pd.DataFrame, col_name: str) -> pd.DataFrame:
+    df = to_pandas(df)
     if not pd.api.types.is_datetime64_any_dtype(df[col_name]):
         df = df.copy()
         df[col_name] = pd.to_datetime(df[col_name])
@@ -236,6 +240,7 @@ def cross_validation(
                     - prediction intervals if `level` is specified.
                     - quantile forecasts if `quantiles` is specified.
         """
+        df = to_pandas(df)
         freq = self._maybe_infer_freq(df, freq)
         df = maybe_convert_col_to_datetime(df, "ds")
         # mlforecast cv code
@@ -344,6 +349,7 @@ def detect_anomalies(
                         an anomaly is defined as a value that is outside of the
                         prediction interval (True or False).
         """
+        df = to_pandas(df)
         freq = self._maybe_infer_freq(df, freq)
         df = maybe_convert_col_to_datetime(df, "ds")
         if h is None:
diff --git a/timecopilot/models/utils/gluonts_forecaster.py b/timecopilot/models/utils/gluonts_forecaster.py
index 25bc2d53..583e03ad 100644
--- a/timecopilot/models/utils/gluonts_forecaster.py
+++ b/timecopilot/models/utils/gluonts_forecaster.py
@@ -12,6 +12,7 @@
 from tqdm import tqdm
 
 from .forecaster import Forecaster, QuantileConverter
+from ..utils.df_utils import to_pandas
 
 
 def fix_freq(freq: str) -> str:
@@ -163,6 +164,7 @@ def forecast(
                 For multi-series data, the output retains the same unique
                 identifiers as the input DataFrame.
         """
+        df = to_pandas(df)
         df = maybe_convert_col_to_float32(df, "y")
         freq = self._maybe_infer_freq(df, freq)
         qc = QuantileConverter(level=level, quantiles=quantiles)
diff --git a/timecopilot/models/utils/parallel_forecaster.py b/timecopilot/models/utils/parallel_forecaster.py
index 4873f3d5..6730ddbd 100644
--- a/timecopilot/models/utils/parallel_forecaster.py
+++ b/timecopilot/models/utils/parallel_forecaster.py
@@ -5,6 +5,7 @@
 import pandas as pd
 
 from .forecaster import Forecaster
+from ..utils.df_utils import to_pandas
 
 
 class ParallelForecaster(Forecaster):
@@ -102,6 +103,7 @@ def forecast(
                 For multi-series data, the output retains the same unique
                 identifiers as the input DataFrame.
         """
+        df = to_pandas(df)
         freq = self._maybe_infer_freq(df, freq)
         fcst_df = self._apply_parallel(
             df.groupby("unique_id"),
diff --git a/timecopilot/utils/df_utils.py b/timecopilot/utils/df_utils.py
new file mode 100644
index 00000000..769b02c2
--- /dev/null
+++ b/timecopilot/utils/df_utils.py
@@ -0,0 +1,39 @@
+from __future__ import annotations
+
+from typing import Any, overload
+
+import pandas as pd
+
+import narwhals as nw
+
+_UNSUPPORTED_DF_MESSAGE = (
+    "Unsupported dataframe type. Install narwhals to enable support for "
+    "polars and other dataframe libraries."
+)
+
+
+@overload
+def to_pandas(df: pd.DataFrame) -> pd.DataFrame: ...
+
+
+@overload
+def to_pandas(df: Any, *, fallback: pd.DataFrame) -> pd.DataFrame: ...
+
+
+@overload
+def to_pandas(df: Any) -> pd.DataFrame: ...
+
+
+def to_pandas(
+    df: Any,
+    *,
+    fallback: pd.DataFrame | None = None,
+) -> pd.DataFrame:
+    if isinstance(df, pd.DataFrame):
+        return df
+    try:
+        return nw.from_native(df).to_pandas()
+    except TypeError as exc:
+        if fallback is not None:
+            return fallback
+        raise TypeError(_UNSUPPORTED_DF_MESSAGE) from exc
diff --git a/timecopilot/utils/experiment_handler.py b/timecopilot/utils/experiment_handler.py
index 5e9bc4cb..46ca673d 100644
--- a/timecopilot/utils/experiment_handler.py
+++ b/timecopilot/utils/experiment_handler.py
@@ -17,6 +17,7 @@
     maybe_convert_col_to_datetime,
     maybe_infer_freq,
 )
+from .df_utils import to_pandas
 
 warnings.simplefilter(
     action="ignore",
@@ -103,6 +104,8 @@ def read_df(path: str | Path) -> pd.DataFrame:
     def _validate_df(df: pd.DataFrame | str | Path) -> pd.DataFrame:
         if isinstance(df, str | Path):
             df = ExperimentDatasetParser.read_df(df)
+        else:
+            df = to_pandas(df)
         if "unique_id" not in df.columns:
             df["unique_id"] = "series_0"
         return maybe_convert_col_to_datetime(df, "ds")
diff --git a/uv.lock b/uv.lock
index 5398584d..33fcb50a 100644
--- a/uv.lock
+++ b/uv.lock
@@ -6830,6 +6830,7 @@ dependencies = [
     { name = "lightgbm" },
     { name = "logfire" },
     { name = "mlforecast" },
+    { name = "narwhals" },
     { name = "neuralforecast" },
     { name = "nixtla" },
     { name = "openai" },
@@ -6883,6 +6884,7 @@ requires-dist = [
     { name = "lightgbm", specifier = ">=4.6.0" },
     { name = "logfire", specifier = ">=4.7.0" },
     { name = "mlforecast", specifier = ">=1.0.2" },
+    { name = "narwhals", specifier = ">=2.7.0" },
     { name = "neuralforecast", specifier = ">=3.0.2" },
     { name = "nixtla", specifier = ">=0.6.6" },
     { name = "openai", specifier = ">=1.99.7" },