diff --git a/src/TensorFlowNET.Core/Eager/Context.cs b/src/TensorFlowNET.Core/Eager/Context.cs
new file mode 100644
index 00000000..b922c6ae
--- /dev/null
+++ b/src/TensorFlowNET.Core/Eager/Context.cs
@@ -0,0 +1,16 @@
+using System;
+using System.Collections.Generic;
+using System.Text;
+
+namespace TensorFlowNET.Core.Eager
+{
+    public class Context
+    {
+        public static int GRAPH_MODE = 0;
+        public static int EAGER_MODE = 1;
+
+        public int default_execution_mode;
+
+
+    }
+}
diff --git a/src/TensorFlowNET.Core/Operation.cs b/src/TensorFlowNET.Core/Operation.cs
index 9b7d9ff2..76d3c815 100644
--- a/src/TensorFlowNET.Core/Operation.cs
+++ b/src/TensorFlowNET.Core/Operation.cs
@@ -17,6 +17,17 @@ namespace TensorFlowNET.Core
         public Tensor[] outputs => _outputs;
         public Tensor[] inputs;
 
+        public Operation(Graph g, string opType, string oper_name)
+        {
+            _graph = g;
+
+            var status = new Status();
+
+            var desc = c_api.TF_NewOperation(g.Handle, opType, oper_name);
+            c_api.TF_SetAttrType(desc, "dtype", DataType.DtInt32);
+            c_api.TF_FinishOperation(desc, status.Handle);
+        }
+
         public Operation(NodeDef node_def, Graph g, object inputs = null, TF_DataType[] output_types = null, object control_inputs = null, TF_DataType[] input_types = null, string original_op = "", OpDef op_def = null)
         {
             _graph = g;
diff --git a/src/TensorFlowNET.Core/tf.cs b/src/TensorFlowNET.Core/tf.cs
index 83bbfc29..60307d04 100644
--- a/src/TensorFlowNET.Core/tf.cs
+++ b/src/TensorFlowNET.Core/tf.cs
@@ -5,6 +5,7 @@ using System.Text;
 using TF_DataType = Tensorflow.DataType;
 using attr_value_pb2 = Tensorflow;
 using Tensorflow;
+using TensorFlowNET.Core.Eager;
 
 namespace TensorFlowNET.Core
 {
@@ -12,11 +13,25 @@ namespace TensorFlowNET.Core
     {
         public static DataType float32 = DataType.DtFloat;
 
+        public static Context context = new Context();
+
+        public static Graph g = new Graph(c_api.TF_NewGraph());
+
         public delegate void Deallocator(IntPtr data, IntPtr size, IntPtr deallocatorData);
 
+        public static unsafe Tensor add(Tensor a, Tensor b)
+        {
+            return null;
+        }
+
         public static unsafe Tensor placeholder(DataType dtype, TensorShape shape = null)
         {
-            return gen_array_ops.placeholder(dtype, shape);
+            var g = ops.get_default_graph();
+            var op = new Operation(g, "Placeholder", "feed");
+
+            var tensor = new Tensor(op, 0, dtype);
+            //return gen_array_ops.placeholder(dtype, shape);
+            return tensor;
         }
 
         public static unsafe Tensor constant(object value)
@@ -38,6 +53,11 @@ namespace TensorFlowNET.Core
             return const_tensor;
         }
 
+        public static void enable_eager_execution()
+        {
+            context.default_execution_mode = Context.EAGER_MODE;
+        }
+
         public static Deallocator FreeTensorDataDelegate = FreeTensorData;
 
         [MonoPInvokeCallback(typeof(Deallocator))]
@@ -55,7 +75,6 @@ namespace TensorFlowNET.Core
 
         public static Graph Graph()
         {
-            Graph g = new Graph(c_api.TF_NewGraph());
             return g;
         }
 
diff --git a/test/TensorFlowNET.UnitTest/OperationsTest.cs b/test/TensorFlowNET.UnitTest/OperationsTest.cs
index ce58f81a..cb6f3ddd 100644
--- a/test/TensorFlowNET.UnitTest/OperationsTest.cs
+++ b/test/TensorFlowNET.UnitTest/OperationsTest.cs
@@ -18,7 +18,18 @@ namespace TensorFlowNET.UnitTest
         [TestMethod]
         public void placeholder()
         {
-            var x = tf.placeholder(tf.float32, shape: new TensorShape(1024, 1024));
+            var x = tf.placeholder(tf.float32);
+        }
+
+        [TestMethod]
+        public void add()
+        {
+            var a = tf.placeholder(tf.float32);
+            var b = tf.placeholder(tf.float32);
+            var c = tf.add(a, b);
+
+            //sess.run(adder_node, { a: 3, b: 4.5})
+            //sess.run(adder_node, {a: [1,3], b: [2, 4]})
         }
     }
 }