diff --git a/data/pipelines/demo.yaml b/data/pipelines/demo.yaml
new file mode 100644
index 0000000..67a24b5
--- /dev/null
+++ b/data/pipelines/demo.yaml
@@ -0,0 +1,3 @@
+steps:
+- step: Segmentation
+- step: TextRecognition
diff --git a/src/htrflow_core/pipeline/pipeline.py b/src/htrflow_core/pipeline/pipeline.py
new file mode 100644
index 0000000..10a01e4
--- /dev/null
+++ b/src/htrflow_core/pipeline/pipeline.py
@@ -0,0 +1,39 @@
+import logging
+from typing import Sequence
+
+from htrflow_core.pipeline.steps import PipelineStep, auto_import, init_step
+
+
+logger = logging.getLogger(__name__)
+
+
+class Pipeline:
+    def __init__(self, steps: Sequence[PipelineStep]):
+        self.steps = steps
+        validate(self)
+
+    @classmethod
+    def from_config(self, config: dict[str, str]):
+        """Init pipeline from config"""
+        return Pipeline([init_step(step) for step in config["steps"]])
+
+    def run(self, volume):
+        """Run pipeline on volume"""
+        volume = auto_import(volume)
+        for i, step in enumerate(self.steps):
+            logger.info("Running step %s (step %d/%d)", step, i+1, len(self.steps))
+            volume = step.run(volume)
+        return volume
+
+    def metadata(self):
+        return [step.metadata for step in self.steps]
+
+
+def validate(pipeline: Pipeline):
+    steps = [step.__class__ for step in pipeline.steps]
+    for i, step in enumerate(steps):
+        for req_step in step.requires:
+            if req_step not in steps[:i]:
+                raise RuntimeError(f"Not valid pipeline: {step.__name__} must be preceded by {req_step.__name__}")
+            logger.info("Validating pipeline: %s is preceded by %s - OK", step.__name__, req_step.__name__)
+    logger.info("Pipeline passed validation")
diff --git a/src/htrflow_core/pipeline/steps.py b/src/htrflow_core/pipeline/steps.py
new file mode 100644
index 0000000..7c25b96
--- /dev/null
+++ b/src/htrflow_core/pipeline/steps.py
@@ -0,0 +1,95 @@
+import logging
+import os
+
+from htrflow_core.dummies.dummy_models import RecognitionModel, SegmentationModel, simple_word_segmentation
+from htrflow_core.volume.volume import Volume
+
+
+logger = logging.getLogger(__name__)
+
+
+class PipelineStep:
+    """Pipeline step base class
+
+    Class attributes:
+        requires: A list of steps that need to precede this step.
+    """
+
+    requires = []
+
+    def run(self, volume: Volume) -> Volume:
+        """Run step"""
+
+    def __str__(self):
+        return f"{self.__class__.__name__}"
+
+
+class Inference(PipelineStep):
+
+    def __init__(self, model):
+        self.model = model
+
+    def run(self, volume):
+        result = self.model(volume.segments())
+        volume.update(result)
+        return volume
+
+
+class Segmentation(Inference):
+    def __init__(self):
+        super().__init__(SegmentationModel())
+
+
+class TextRecognition(Inference):
+    def __init__(self):
+        super().__init__(RecognitionModel())
+
+
+class WordSegmentation(PipelineStep):
+
+    requires = [TextRecognition]
+
+    def run(self, volume):
+        results = simple_word_segmentation(volume.leaves())
+        volume.update(results)
+        return volume
+
+
+class ExportStep(PipelineStep):
+    def __init__(self, dest, serializer):
+        self.dest = dest
+        self.serializer = serializer
+
+    def run(self, volume):
+        volume.save(self.dest, self.serializer)
+        return volume
+
+
+def auto_import(source) -> Volume:
+    """Import volume from `source`
+
+    Automatically detects import type from the input. Supported types
+    are:
+        - directories with images
+    """
+    if isinstance(source, Volume):
+        return source
+    elif isinstance(source, str):
+        if os.path.isdir(source):
+            logger.info("Loading volume from directory %s", source)
+            return Volume.from_directory(source)
+    raise ValueError(f"Could not infer import type for '{source}'")
+
+
+def all_subclasses(cls):
+    return set(cls.__subclasses__()).union([s for c in cls.__subclasses__() for s in all_subclasses(c)])
+
+# Mapping class name -> class
+# Ex. {segmentation: `steps.Segmentation`}
+STEPS = {cls_.__name__.lower(): cls_ for cls_ in all_subclasses(PipelineStep)}
+
+
+def init_step(step):
+    name = step["step"].lower()
+    kwargs = step.get("settings", {})
+    return STEPS[name](**kwargs)
diff --git a/src/main.py b/src/main.py
new file mode 100644
index 0000000..7ec819d
--- /dev/null
+++ b/src/main.py
@@ -0,0 +1,25 @@
+"""
+Entrypoint for HTRFLOW
+
+Usage:
+> python src/main.py <pipeline.yaml> <input_directory>
+
+Example:
+> python src/main.py data/pipelines/demo.yaml data/demo_images/A0068699
+"""
+import logging
+import sys
+
+import yaml
+
+from htrflow_core.pipeline.pipeline import Pipeline
+
+
+logging.basicConfig(filename='htrflow.log', level=logging.INFO, filemode='w')
+
+if __name__ == '__main__':
+    with open(sys.argv[1], 'r') as f:
+        config = yaml.safe_load(f)
+
+    pipe = Pipeline.from_config(config)
+    pipe.run("data/demo_images/A0068699")