diff --git a/app/playground/main.cpp b/app/playground/main.cpp
index b370892..cd9e742 100644
--- a/app/playground/main.cpp
+++ b/app/playground/main.cpp
@@ -10,10 +10,27 @@
 int main() {
   using namespace pls::dataflow;
 
-  graph<inputs<int, int>, outputs<int, int>, 8> graph;
-  graph.output<0>().connect(graph.input<0>());
-  graph.output<1>().connect(graph.input<1>());
+  graph<inputs<int, int>, outputs<int, int>, 2> graph2;
+  graph2.input<0>() >> graph2.output<1>();
+  graph2.input<1>() >> graph2.output<0>();
 
-  graph.input<0>().push_token(pls::dataflow::internal::token<int>());
-  graph.input<1>().push_token(pls::dataflow::internal::token<int>());
+  graph<inputs<int, int>, outputs<int, int>, 2> graph1;
+  graph1.input<0>() >> graph2.external_input<0>();
+  graph1.input<1>() >> graph2.external_input<1>();
+  graph2.external_output<0>() >> graph1.output<0>();
+  graph2.external_output<1>() >> graph1.output<1>();
+
+  graph1.push_input(1, 2);
+  graph1.push_input(3, 4);
+
+  auto result1 = graph1.get_output();
+  std::cout << std::get<0>(result1) << ", " << std::get<1>(result1) << std::endl;
+
+  graph1.push_input(5, 6);
+
+  auto result2 = graph1.get_output();
+  std::cout << std::get<0>(result2) << ", " << std::get<1>(result2) << std::endl;
+
+  auto result3 = graph1.get_output();
+  std::cout << std::get<0>(result3) << ", " << std::get<1>(result3) << std::endl;
 }
diff --git a/lib/pls/CMakeLists.txt b/lib/pls/CMakeLists.txt
index 45b0740..41a974b 100644
--- a/lib/pls/CMakeLists.txt
+++ b/lib/pls/CMakeLists.txt
@@ -47,7 +47,7 @@ add_library(pls STATIC
         include/pls/internal/scheduling/scheduler_impl.h
         include/pls/internal/scheduling/task.h src/internal/scheduling/task.cpp
         include/pls/internal/scheduling/scheduler_memory.h src/internal/scheduling/scheduler_memory.cpp
-        include/pls/internal/scheduling/lambda_task.h include/pls/dataflow/inputs.h include/pls/dataflow/outputs.h)
+        include/pls/internal/scheduling/lambda_task.h include/pls/dataflow/inputs.h include/pls/dataflow/outputs.h include/pls/dataflow/input.h include/pls/dataflow/output.h)
 # Add everything in `./include` to be in the include path of this project
 target_include_directories(pls
         PUBLIC
diff --git a/lib/pls/include/pls/dataflow/graph.h b/lib/pls/include/pls/dataflow/graph.h
index 46fab38..73b4e23 100644
--- a/lib/pls/include/pls/dataflow/graph.h
+++ b/lib/pls/include/pls/dataflow/graph.h
@@ -4,27 +4,172 @@
 
 #include "inputs.h"
 #include "outputs.h"
+#include "input.h"
+#include "output.h"
+
+#include "internal/token.h"
 
 namespace pls {
 namespace dataflow {
 
 template<typename I, typename O, int P>
 class graph {
-  using internal_inputs = typename I::template internal_inputs<P>;
-  using internal_outputs = typename O::template internal_outputs<P>;
+  /**
+   * Boilerplate for callbacks of dataflow input/output handlers.
+   */
+  struct external_input_cb {
+    graph *graph_;
+
+    void one_input(int pos, internal::token_color color) {
+      graph_->one_external_input(pos, color);
+    }
+    void all_inputs(internal::token_color color) {
+      graph_->all_external_inputs(color);
+    }
+  };
+  struct internal_output_cb {
+    graph *graph_;
+
+    void one_input(int pos, internal::token_color color) {
+      graph_->one_internal_output(pos, color);
+    }
+    void all_inputs(internal::token_color color) {
+      graph_->all_internal_outputs(color);
+    }
+  };
+
+  /**
+   * Handling of buffers for dataflow.
+   */
+  using exeternal_inputs = typename I::template internal_inputs<P, graph<I, O, P>::external_input_cb>;
+  using external_outputs = typename O::template internal_outputs<P>;
+  exeternal_inputs external_inputs_{external_input_cb{this}};
+  external_outputs external_outputs_{};
+
+  void one_external_input(int pos, internal::token_color color) {
+    // We do currently not support 'single external inputs'.
+    // This prevents some fine-grained parallelism, but eases implementation
+  }
+  void all_external_inputs(internal::token_color external_color) {
+    auto my_clock = input_clock_++;
+    internal::token_color internal_color{my_clock, external_color.depth_ + 1};
+    auto my_index = internal_color.get_index(P);
+
+    while (output_clear_[my_index] != clear) {
+      PLS_ERROR("Full inputs are not implemented jet. Place waiting code in here/yield to scheduler.")
+    }
+
+    output_clear_[my_index] = waiting;
+    input_output_color_matching_[my_index] = {external_color, internal_color};
+    external_inputs_.push_to_outputs(internal_inputs_, external_color.get_index(P), internal_color);
+  }
+
+  using internal_outputs = typename I::template internal_inputs<P, graph<I, O, P>::internal_output_cb>;
+  using internal_inputs = typename O::template internal_outputs<P>;
+  internal_outputs internal_outputs_{internal_output_cb{this}};
+  internal_inputs internal_inputs_{};
+
+  void one_internal_output(int pos, internal::token_color color) {
+    // We do not support 'single flow' output. Every graph is well behaved, e.g. will emmit exactly one output tuple.
+  }
+  void all_internal_outputs(internal::token_color color) {
+    auto color_matching = input_output_color_matching_[color.get_index(P)];
+    auto external_color = std::get<0>(color_matching);
+    auto internal_color = std::get<1>(color_matching);
+    if (external_color == internal_color) {
+      output_clear_[color.get_index(P)] = finished;
+    } else {
+      internal_outputs_.push_to_outputs(external_outputs_, internal_color.get_index(P), external_color);
+      output_clear_[color.get_index(P)] = clear;
+    }
+  }
+
+  /**
+   * Handling of colors/maximum concurrent tokens in the system.
+   */
+  std::atomic<unsigned int> input_clock_{0};
+  std::atomic<unsigned int> output_clock_{0};
+  std::array<std::pair<internal::token_color, internal::token_color>, P> input_output_color_matching_{};
+  enum output_state { clear, waiting, finished };
+  std::array<output_state, P> output_clear_{clear};
 
  public:
-  internal_inputs inputs_;
-  internal_outputs outputs_;
+  template<int N>
+  using external_output_at = output<P, typename external_outputs::template raw_type_at<N>>;
+  template<int N>
+  using external_input_at = input<P, typename exeternal_inputs::template raw_type_at<N>>;
 
   template<int N>
-  decltype(outputs_.template get<N>()) output() {
-    return outputs_.template get<N>();
+  external_output_at<N> external_output() {
+    external_output_at<N> o{external_outputs_.template get<N>()};
+    return o;
+  }
+  template<int N>
+  external_input_at<N> external_input() {
+    external_input_at<N> i{external_inputs_.template get<N>()};
+    return i;
   }
 
   template<int N>
-  decltype(inputs_.template get<N>()) input() {
-    return inputs_.template get<N>();
+  using input_at = output<P, typename internal_inputs::template raw_type_at<N>>;
+  template<int N>
+  using output_at = input<P, typename internal_outputs::template raw_type_at<N>>;
+
+  template<int N>
+  output_at<N> output() {
+    output_at<N> o{internal_outputs_.template get<N>()};
+    return o;
+  }
+  template<int N>
+  input_at<N> input() {
+    input_at<N> i{internal_inputs_.template get<N>()};
+    return i;
+  }
+
+  template<typename ...ARGS>
+  void push_input(ARGS ...values) {
+    auto my_clock = input_clock_++;
+    internal::token_color color{my_clock, 0};
+    auto my_index = color.get_index(P);
+
+    while (output_clear_[my_index] != clear) {
+      PLS_ERROR("Full inputs are not implemented jet. Place waiting code in here/yield to scheduler.")
+    }
+
+    output_clear_[my_index] = waiting;
+    input_output_color_matching_[my_index] = {color, color};
+
+    push_input_rec<0, ARGS...>(color, values...);
+  }
+  template<int POS, typename ...TAIL>
+  void push_input_rec(internal::token_color color) {}
+  template<int POS, typename HEAD, typename ...TAIL>
+  void push_input_rec(internal::token_color color, HEAD h, TAIL ...t) {
+    internal_inputs_.template get<POS>().push_token(internal::token<HEAD>{h, color});
+    push_input_rec<POS + 1, TAIL...>(color, t...);
+  }
+
+  typename internal_outputs::raw_types get_output() {
+    while (true) {
+      if (output_clock_ > input_clock_) {
+        PLS_ERROR("Must not try to pull outputs if there are not enough inputs.")
+      }
+
+      auto color_matching = input_output_color_matching_[output_clock_ % P];
+      if (std::get<0>(color_matching) == std::get<1>(color_matching)) {
+        break;
+      }
+      output_clock_++;
+    }
+
+    while (output_clear_[output_clock_ % P] != finished) {
+      PLS_ERROR("Blocking/Unfinishe outputs are not implemented jet. Place waiting code in here/yield to scheduler.")
+    }
+
+    auto result = internal_outputs_.get_outputs(output_clock_ % P);
+    output_clear_[output_clock_ % P] = clear;
+    output_clock_++;
+    return result;
   }
 };
 
diff --git a/lib/pls/include/pls/dataflow/inputs.h b/lib/pls/include/pls/dataflow/inputs.h
index c7d453e..8e397fa 100644
--- a/lib/pls/include/pls/dataflow/inputs.h
+++ b/lib/pls/include/pls/dataflow/inputs.h
@@ -11,8 +11,8 @@ namespace dataflow {
 
 template<typename I1, typename ...I>
 struct inputs {
-  template<int P>
-  using internal_inputs = internal::inputs<P, I1, I...>;
+  template<int P, typename CB>
+  using internal_inputs = internal::inputs<P, CB, I1, I...>;
 };
 
 }
diff --git a/lib/pls/include/pls/dataflow/internal/input.h b/lib/pls/include/pls/dataflow/internal/input.h
index 889351f..c5b0f97 100644
--- a/lib/pls/include/pls/dataflow/internal/input.h
+++ b/lib/pls/include/pls/dataflow/internal/input.h
@@ -37,11 +37,17 @@ class input {
   }
 
  public:
+  input() = default;
+
   void push_token(token<T> token) {
     tokens_[token.color().get_index(P)] = token;
     cb_->token_pushed(input_pos_, token.color());
   }
 
+  token<T> token_at(int p) {
+    return tokens_[p];
+  }
+
   void set_cb(push_token_cb *cb, int input_pos) {
     cb_ = cb;
     input_pos_ = input_pos;
diff --git a/lib/pls/include/pls/dataflow/internal/inputs.h b/lib/pls/include/pls/dataflow/internal/inputs.h
index 5c2218b..100665d 100644
--- a/lib/pls/include/pls/dataflow/internal/inputs.h
+++ b/lib/pls/include/pls/dataflow/internal/inputs.h
@@ -6,31 +6,50 @@
 #include <array>
 
 #include "input.h"
+#include "outputs.h"
 
 namespace pls {
 namespace dataflow {
 namespace internal {
 
-template<int P, typename ...I>
+template<int P, typename CB, typename ...I>
 class inputs : push_token_cb {
+ public:
+  using raw_types = std::tuple<I...>;
   using values_type = std::tuple<input<P, I>...>;
-  values_type values_;
+
+  template<int N>
+  using raw_type_at = typename std::tuple_element<N, raw_types>::type;
+  template<int N>
+  using value_type_at = typename std::tuple_element<N, values_type>::type;
   static constexpr unsigned int num_values = std::tuple_size<values_type>::value;
 
+ private:
+  values_type values_;
   std::array<std::atomic<unsigned int>, P> required_inputs_;
+  CB cb_;
 
  public:
-  inputs() {
+  explicit inputs(CB cb) : cb_{cb} {
     for (int i = 0; i < P; i++) {
       required_inputs_[i] = num_values;
     }
     init_cb<0, I...>::call(this);
   }
+  inputs(inputs &&other) = delete;
+  inputs(const inputs &other) = delete;
+  inputs &operator=(inputs &&other) = delete;
+  inputs &operator=(const inputs &other) = delete;
+
+  void token_pushed(int pos, token_color color) override {
+    auto index = color.get_index(P);
+    int current_required = --required_inputs_[index];
+
+    cb_.one_input(pos, color);
 
-  void token_pushed(int /*pos*/, token_color color) override {
-    int current_required = --required_inputs_[color.get_index(P)];
     if (current_required == 0) {
-      std::cout << "All Inputs Avaliable" << std::endl; // TODO: Add proper callback in here
+      cb_.all_inputs(color);
+      required_inputs_[index] = num_values;
     }
   }
 
@@ -39,14 +58,47 @@ class inputs : push_token_cb {
     return std::get<N>(values_);
   }
 
+  raw_types get_outputs(int p) {
+    return fill_output<0, I...>::call(*this, p);
+  }
+  template<int POS, typename ...TAIL>
+  struct fill_output {
+    static std::tuple<> call(inputs<P, CB, I...> &self, int p) {
+      return {};
+    }
+  };
+  template<int POS, typename HEAD, typename ...TAIL>
+  struct fill_output<POS, HEAD, TAIL...> {
+    static std::tuple<HEAD, TAIL...> call(inputs<P, CB, I...> &self, int p) {
+      std::tuple<decltype(self.get<POS>().token_at(p).value())> our_tuple{self.get<POS>().token_at(p).value()};
+      return std::tuple_cat(our_tuple, fill_output<POS + 1, TAIL...>::call(self, p));
+    }
+  };
+
+  void push_to_outputs(outputs<P, I...> &outputs, int input_index, token_color new_color) {
+    push_to_outputs_rec<0, I...>::call(this, &outputs, input_index, new_color);
+  }
+  template<int POS, typename ...TAIL>
+  struct push_to_outputs_rec {
+    static void call(inputs<P, CB, I...> *inputs, outputs<P, I...> *outputs, int input_index, token_color new_color) {}
+  };
+  template<int POS, typename HEAD, typename ...TAIL>
+  struct push_to_outputs_rec<POS, HEAD, TAIL...> {
+    static void call(inputs<P, CB, I...> *inputs, outputs<P, I...> *outputs, int input_index, token_color new_color) {
+      auto token = inputs->get<POS>().token_at(input_index);
+      outputs->template get<POS>().push_token({token.value(), new_color});
+      push_to_outputs_rec<POS + 1, TAIL...>::call(inputs, outputs, input_index, new_color);
+    }
+  };
+
 // TODO: Change CB code using proper templating to save method calls during execution...
   template<int POS, typename ...TAIL>
   struct init_cb {
-    static void call(inputs<P, I...> *inputs) { }
+    static void call(inputs<P, CB, I...> *inputs) {}
   };
   template<int POS, typename HEAD, typename ...TAIL>
   struct init_cb<POS, HEAD, TAIL...> {
-    static void call(inputs<P, I...> *inputs) {
+    static void call(inputs<P, CB, I...> *inputs) {
       inputs->get<POS>().set_cb(inputs, POS);
       init_cb<POS + 1, TAIL...>::call(inputs);
     }
diff --git a/lib/pls/include/pls/dataflow/internal/outputs.h b/lib/pls/include/pls/dataflow/internal/outputs.h
index ca616ea..77ae3c1 100644
--- a/lib/pls/include/pls/dataflow/internal/outputs.h
+++ b/lib/pls/include/pls/dataflow/internal/outputs.h
@@ -12,12 +12,21 @@ namespace internal {
 
 template<int P, typename ...O>
 class outputs {
+ public:
+  using raw_types = std::tuple<O...>;
   using values_type = std::tuple<output<P, O>...>;
+
+  template<int N>
+  using raw_type_at = typename std::tuple_element<N, raw_types>::type;
+  template<int N>
+  using value_type_at = typename std::tuple_element<N, values_type>::type;
+
+ private:
   values_type values_;
 
  public:
   template<int N>
-  typename std::tuple_element<N, values_type>::type &get() {
+  value_type_at<N> &get() {
     return std::get<N>(values_);
   }
 };
diff --git a/lib/pls/include/pls/dataflow/internal/token.h b/lib/pls/include/pls/dataflow/internal/token.h
index a6243f6..edb36b6 100644
--- a/lib/pls/include/pls/dataflow/internal/token.h
+++ b/lib/pls/include/pls/dataflow/internal/token.h
@@ -13,10 +13,15 @@ namespace internal {
  */
 struct token_color {
   unsigned int clock_;
+  unsigned int depth_;
 
   int get_index(int parallel_limit) const {
     return clock_ % parallel_limit;
   }
+
+  bool operator==(const token_color &other) {
+    return other.clock_ == clock_ && other.depth_ == depth_;
+  }
 };
 
 template<typename T>
@@ -25,6 +30,9 @@ class token {
   token_color color_;
 
  public:
+  token() : color_{} {}; // Default Constructor Stays uninitialized
+  token(T value, token_color color) : value_{value}, color_{color} {};
+
   T value() const { return value_; }
   token_color color() const { return color_; }
 };