diff --git a/pandas/core/apply.py b/pandas/core/apply.py
index bb39e18caeaa2..968244b581826 100644
--- a/pandas/core/apply.py
+++ b/pandas/core/apply.py
@@ -538,8 +538,21 @@ def normalize_dictlike_arg(
             raise SpecificationError("nested renamer is not supported")
 
         if obj.ndim != 1:
+
+            # Check if it is an aggregate over multiple columns,
+            #   this means a tuple of columns has been passed,
+            #   so we need to unnest to check fo non existing columns.
+            func_keys = set()
+            for func_key in func.keys():
+                if is_list_like(func_key):
+                    # a tuple of columns has been passed, we unnest.
+                    for key in func_key:
+                        func_keys.add(key)
+                else:
+                    func_keys.add(func_key)
+
             # Check for missing columns on a frame
-            cols = set(func.keys()) - set(obj.columns)
+            cols = func_keys - set(obj.columns)
             if len(cols) > 0:
                 cols_sorted = list(safe_sort(list(cols)))
                 raise KeyError(f"Column(s) {cols_sorted} do not exist")
diff --git a/pandas/core/frame.py b/pandas/core/frame.py
index d9264e465dbef..783983f2243b9 100644
--- a/pandas/core/frame.py
+++ b/pandas/core/frame.py
@@ -3500,7 +3500,9 @@ def __getitem__(self, key):
 
         # We are left with two options: a single key, and a collection of keys,
         # We interpret tuples as collections only for non-MultiIndex
-        is_single_key = isinstance(key, tuple) or not is_list_like(key)
+        is_single_key = (
+            isinstance(key, tuple) or not is_list_like(key)
+        ) and self.columns.nlevels > 1
 
         if is_single_key:
             if self.columns.nlevels > 1:
diff --git a/pandas/tests/groupby/aggregate/test_aggregate.py b/pandas/tests/groupby/aggregate/test_aggregate.py
index 0a693967fbb19..53d3492ea0039 100644
--- a/pandas/tests/groupby/aggregate/test_aggregate.py
+++ b/pandas/tests/groupby/aggregate/test_aggregate.py
@@ -1274,3 +1274,15 @@ def func(ser):
 
     expected = DataFrame([[1.0]], index=[1])
     tm.assert_frame_equal(res, expected)
+
+
+def test_named_agg_multiple_columns():
+    # GH29268
+    df = DataFrame({"a": [5, 7, 9, 11], "b": [8, 23, 5, 9]})
+    df["group"] = [0, 0, 1, 1]
+
+    result = df.groupby("group").agg(
+        diff_a_b=(("a", "b"), lambda x: x["a"].max() - x["b"].max())
+    )
+    expected = DataFrame({"diff_a_b": [16, 2]}, index=pd.Index([0, 1], name="group"))
+    tm.assert_frame_equal(result, expected)