diff --git a/README.md b/README.md
index daca471..285df1b 100644
--- a/README.md
+++ b/README.md
@@ -1,2 +1,16 @@
 # projekt-sztuczna-inteligencja
 
+## Uruchomienie projektu
+
+Wymaga Pythona oraz pip.
+
+Należy zainstalować zależności:
+```sh
+pip install -r requirements.txt
+```
+
+A następnie uruchomić serwer
+```sh
+cd src
+python3 main.py
+```
diff --git a/requirements.txt b/requirements.txt
new file mode 100644
index 0000000..7ba0612
--- /dev/null
+++ b/requirements.txt
@@ -0,0 +1,5 @@
+Flask==2.0.1
+matplotlib==3.4.2
+numpy==1.20.3
+scikit-learn==0.24.2
+scipy==1.6.3
diff --git a/src/decision_tree/main.py b/src/decision_tree.py
similarity index 62%
rename from src/decision_tree/main.py
rename to src/decision_tree.py
index 6c3e7de..9b1031b 100755
--- a/src/decision_tree/main.py
+++ b/src/decision_tree.py
@@ -1,31 +1,31 @@
 #!/usr/bin/python3
 from sklearn import tree
 from pprint import PrettyPrinter
+import pickle
+import os
 
 pp = PrettyPrinter(indent=2, compact=True)
 def p(*args, **kwargs):
   pp.pprint(*args, **kwargs)
 
-def invoke_consume_exceptions(function, *args, **kwargs):
-  try:
-    return function(*args, **kwargs)
-  except:
-    return None
-
 def read_tsv_from(filename):
   from csv import reader
   with open(filename, 'r') as f:
     header, *rows = list(reader(f, delimiter='\t'))
     return [dict(zip(header, (el.strip() for el in row))) for row in rows]
 
-def main():
-  from sys import argv
-  import os
-  import pathlib
-  source_file = argv[1]
-  invoke_consume_exceptions(os.mkdir, os.path.dirname(source_file))
-  data = read_tsv_from(source_file)
+def serialize(clf):
+  with open('./data/tree', 'wb') as f:
+    pickle.dump(clf, f)
 
+def deserialize():
+  with open('./data/tree', 'rb') as f:
+    return pickle.load(f)
+
+def init(source_file, cache=False):
+  from random import randint
+  defaults, *data = read_tsv_from(source_file)
+  defaults.pop("nazwa")
   types = dict()
   for row in data:
     for (key, value) in row.items():
@@ -42,18 +42,29 @@ def main():
   t2n = dict((key, dict(v)) for (key, v) in base.items())
   n2t = dict((key, dict((b, a) for (a, b) in v)) for (key, v) in base.items())
 
+
   X = [[
-    t2n[name][feature] 
+    t2n[name][feature]
     for (name, feature) in sample.items() 
     if name not in ['nazwa', 'polka']] for sample in data if 'polka' in sample]
   
   Y = [t2n['polka'][sample['polka']] for sample in data if 'polka' in sample]
   
-  clf = tree.DecisionTreeClassifier()
-  clf.fit(X, Y)
-  l = clf.get_n_leaves()
-  d = clf.get_depth()
-  print(f'Leaves: {l}\nDepth: {d}')
+  if cache and os.path.isfile('./data/tree'):
+    clf = deserialize()
+  else:
+    clf = tree.DecisionTreeClassifier()
+    clf.fit(X, Y)
+    if cache:
+      serialize(clf)
+
+  return [
+    types,
+    t2n,
+    n2t,
+    defaults,
+    clf
+  ]
 
 if __name__ == '__main__':
-  main()
\ No newline at end of file
+  init('./data/productsTree.tsv')
diff --git a/src/main.py b/src/main.py
new file mode 100644
index 0000000..ebb4a4f
--- /dev/null
+++ b/src/main.py
@@ -0,0 +1,40 @@
+from flask import Flask, redirect, jsonify, request
+import decision_tree
+
+app = Flask(
+  __name__,
+  static_url_path='',
+  static_folder='.')
+
+@app.route('/api/types/', defaults={ 'searched_type': None })
+@app.route('/api/types/<searched_type>')
+def api_get_features_types(searched_type):
+  return t2n[searched_type] if searched_type else t2n
+
+@app.route('/api/types/defaults')
+def api_get_default_features():
+  return jsonify(defaults)
+
+@app.route('/api/decide', methods=['POST'])
+def api_predict_shelf():
+  json = request.get_json(force=True)
+  defs = dict(defaults)
+  defs.pop('polka')
+  keys = list(json.keys())
+  X = []
+
+  for (key, default) in defs.items():
+    if key in keys:
+      X.append(t2n[key][json[key]])
+    else:
+      X.append(t2n[key][default])
+
+  return jsonify([ n2t['polka'][clf.predict([X]).tolist()[0]] ])
+
+@app.route('/')
+def index():
+  return redirect('/index.html')
+
+categories, t2n, n2t, defaults, clf = decision_tree.init('data/productsTree.tsv', cache=True)
+categories = dict((key, list(vals)) for (key, vals) in categories.items())
+app.run()
diff --git a/test.sh b/test.sh
new file mode 100755
index 0000000..f0d4ba7
--- /dev/null
+++ b/test.sh
@@ -0,0 +1,10 @@
+#!/bin/sh
+
+decide () {
+	echo Input: "$1"
+	echo Output:
+	curl "http://localhost:5000/api/decide" -d"$1"
+}
+
+decide '{ "nazwa": "chleb", "kategoria": "kuchnia", "typ_zywnosci": "gotowe" }'
+decide '{ "nazwa": "parowki", "kategoria": "kuchnia", "typ_zywnosci": "mieso", "termin_przydatnosci": "krotki" }'