diff --git a/jedi/inference/base_value.py b/jedi/inference/base_value.py
index 05134773..4d14f1a9 100644
--- a/jedi/inference/base_value.py
+++ b/jedi/inference/base_value.py
@@ -258,6 +258,7 @@ class Value(HelperValueMixin, BaseValue):
     def _as_context(self):
         raise NotImplementedError('Not all values need to be converted to contexts: %s', self)
 
+    @property
     def name(self):
         raise NotImplementedError
 
diff --git a/jedi/inference/gradual/typing.py b/jedi/inference/gradual/typing.py
index 3b3cdf17..be353391 100644
--- a/jedi/inference/gradual/typing.py
+++ b/jedi/inference/gradual/typing.py
@@ -5,8 +5,9 @@ values.
 
 This file deals with all the typing.py cases.
 """
+import itertools
 from jedi import debug
-from jedi.inference.compiled import builtin_from_name
+from jedi.inference.compiled import builtin_from_name, create_simple_object
 from jedi.inference.base_value import ValueSet, NO_VALUES, Value, \
     LazyValueWrapper
 from jedi.inference.lazy_value import LazyKnownValues
@@ -81,7 +82,8 @@ class TypingModuleName(NameWrapper):
         elif name == 'TypedDict':
             # TODO doesn't even exist in typeshed/typing.py, yet. But will be
             # added soon.
-            pass
+            yield TypedDictBase.create_cached(
+                inference_state, self.parent_context, self.tree_name)
         elif name in ('no_type_check', 'no_type_check_decorator'):
             # This is not necessary, as long as we are not doing type checking.
             for c in self._wrapped_name.infer():  # Fuck my life Python 2
@@ -339,3 +341,74 @@ class CastFunction(BaseTypingValue):
     @repack_with_argument_clinic('type, object, /')
     def py__call__(self, type_value_set, object_value_set):
         return type_value_set.execute_annotation()
+
+
+class TypedDictBase(BaseTypingValue):
+    """
+    This class has no responsibilities and is just here to make sure that typed
+    dicts can be identified.
+    """
+
+
+class TypedDictClass(Value):
+    """
+    This represents a class defined like:
+
+        class Foo(TypedDict):
+            bar: str
+    """
+    def __init__(self, definition_class):
+        super().__init__(definition_class.inference_state, definition_class.parent_context)
+        self.tree_node = definition_class.tree_node
+        self._definition_class = definition_class
+
+    def get_filters(self, origin_scope=None):
+        """
+        A TypedDict doesn't have attributes.
+        """
+        o, = self.inference_state.builtins_module.py__getattribute__('object')
+        return o.get_filters()
+
+    @property
+    def name(self):
+        return ValueName(self, self.tree_node.name)
+
+    def py__call__(self, arguments):
+        return ValueSet({TypedDict(self._definition_class)})
+
+
+class TypedDict(LazyValueWrapper):
+    """Represents the instance version of ``TypedDictClass``."""
+    def __init__(self, definition_class):
+        self.inference_state = definition_class.inference_state
+        self.parent_context = definition_class.parent_context
+        self.tree_node = definition_class.tree_node
+        self._definition_class = definition_class
+
+    @property
+    def name(self):
+        return ValueName(self, self.tree_node.name)
+
+    def py__simple_getitem__(self, index):
+        if isinstance(index, str):
+            return ValueSet.from_sets(
+                name.infer()
+                for filter in self._definition_class.get_filters(is_instance=True)
+                for name in filter.get(index)
+            )
+        return NO_VALUES
+
+    def get_key_values(self):
+        filtered_values = itertools.chain.from_iterable((
+            f.values()
+            for f in self._definition_class.get_filters(is_instance=True)
+        ))
+        return ValueSet({
+            create_simple_object(self.inference_state, v.string_name)
+            for v in filtered_values
+        })
+
+    def _get_wrapped_value(self):
+        d, = self.inference_state.builtins_module.py__getattribute__('dict')
+        result, = d.execute_with_values()
+        return result
diff --git a/jedi/inference/syntax_tree.py b/jedi/inference/syntax_tree.py
index 6411fc81..bc902020 100644
--- a/jedi/inference/syntax_tree.py
+++ b/jedi/inference/syntax_tree.py
@@ -26,6 +26,7 @@ from jedi.inference.compiled.access import COMPARISON_OPERATORS
 from jedi.inference.cache import inference_state_method_cache
 from jedi.inference.gradual.stub_value import VersionInfo
 from jedi.inference.gradual import annotation
+from jedi.inference.gradual.typing import TypedDictClass
 from jedi.inference.names import TreeNameDefinition
 from jedi.inference.context import CompForContext
 from jedi.inference.value.decorator import Decoratee
@@ -748,6 +749,8 @@ def _apply_decorators(context, node):
             parent_context=context,
             tree_node=node
         )
+        if decoratee_value.is_typeddict():
+            decoratee_value = TypedDictClass(decoratee_value)
     else:
         decoratee_value = FunctionValue.from_context(context, node)
     initial = values = ValueSet([decoratee_value])
diff --git a/jedi/inference/value/instance.py b/jedi/inference/value/instance.py
index 2eca9f9d..b884b14d 100644
--- a/jedi/inference/value/instance.py
+++ b/jedi/inference/value/instance.py
@@ -547,10 +547,10 @@ class InstanceClassFilter(AbstractFilter):
         self._class_filter = class_filter
 
     def get(self, name):
-        return self._convert(self._class_filter.get(name, from_instance=True))
+        return self._convert(self._class_filter.get(name))
 
     def values(self):
-        return self._convert(self._class_filter.values(from_instance=True))
+        return self._convert(self._class_filter.values())
 
     def _convert(self, names):
         return [
@@ -586,7 +586,7 @@ class SelfAttributeFilter(ClassFilter):
             if trailer.type == 'trailer' \
                     and len(trailer.parent.children) == 2 \
                     and trailer.children[0] == '.':
-                if name.is_definition() and self._access_possible(name, from_instance=True):
+                if name.is_definition() and self._access_possible(name):
                     # TODO filter non-self assignments instead of this bad
                     #      filter.
                     if self._is_in_right_scope(trailer.parent.children[0], name):
diff --git a/jedi/inference/value/klass.py b/jedi/inference/value/klass.py
index 37b743b3..41c669d7 100644
--- a/jedi/inference/value/klass.py
+++ b/jedi/inference/value/klass.py
@@ -38,11 +38,11 @@ py__doc__()                            Returns the docstring for a value.
 """
 from jedi import debug
 from jedi._compatibility import use_metaclass
-from jedi.parser_utils import get_cached_parent_scope
+from jedi.parser_utils import get_cached_parent_scope, expr_is_dotted
 from jedi.inference.cache import inference_state_method_cache, CachedMetaClass, \
     inference_state_method_generator_cache
 from jedi.inference import compiled
-from jedi.inference.lazy_value import LazyKnownValues
+from jedi.inference.lazy_value import LazyKnownValues, LazyTreeValue
 from jedi.inference.filters import ParserTreeFilter
 from jedi.inference.names import TreeNameDefinition, ValueName
 from jedi.inference.arguments import unpack_arglist, ValuesArguments
@@ -104,12 +104,12 @@ class ClassFilter(ParserTreeFilter):
             node = get_cached_parent_scope(self._used_names, node)
         return False
 
-    def _access_possible(self, name, from_instance=False):
+    def _access_possible(self, name):
         # Filter for ClassVar variables
         # TODO this is not properly done, yet. It just checks for the string
         # ClassVar in the annotation, which can be quite imprecise. If we
         # wanted to do this correct, we would have to infer the ClassVar.
-        if not from_instance:
+        if not self._is_instance:
             expr_stmt = name.get_definition()
             if expr_stmt is not None and expr_stmt.type == 'expr_stmt':
                 annassign = expr_stmt.children[1]
@@ -122,9 +122,9 @@ class ClassFilter(ParserTreeFilter):
         return not name.value.startswith('__') or name.value.endswith('__') \
             or self._equals_origin_scope()
 
-    def _filter(self, names, from_instance=False):
+    def _filter(self, names):
         names = super(ClassFilter, self)._filter(names)
-        return [name for name in names if self._access_possible(name, from_instance)]
+        return [name for name in names if self._access_possible(name)]
 
 
 class ClassMixin(object):
@@ -270,6 +270,36 @@ class ClassValue(use_metaclass(CachedMetaClass, ClassMixin, FunctionAndClassBase
             self.inference_state.builtins_module.py__getattribute__('object')
         )]
 
+    @inference_state_method_cache(default=False)
+    def is_typeddict(self):
+        # TODO Do a proper mro resolution. Currently we are just listing
+        # classes. However, it's a complicated algorithm.
+        from jedi.inference.gradual.typing import TypedDictBase
+        for lazy_cls in self.py__bases__():
+            if not isinstance(lazy_cls, LazyTreeValue):
+                return False
+            tree_node = lazy_cls.data
+            # Only resolve simple classes, stuff like Iterable[str] are more
+            # intensive to resolve and if generics are involved, we know it's
+            # not a TypedDict.
+            if not expr_is_dotted(tree_node):
+                return False
+
+            for cls in lazy_cls.infer():
+                if isinstance(cls, TypedDictBase):
+                    return True
+                try:
+                    method = cls.is_typeddict
+                except AttributeError:
+                    # We're only dealing with simple classes, so just returning
+                    # here should be fine. This only happens with e.g. compiled
+                    # classes.
+                    return False
+                else:
+                    if method():
+                        return True
+        return False
+
     def py__getitem__(self, index_value_set, contextualized_node):
         from jedi.inference.gradual.base import GenericClass
         if not index_value_set:
diff --git a/jedi/parser_utils.py b/jedi/parser_utils.py
index 0f7ba429..7f69a80c 100644
--- a/jedi/parser_utils.py
+++ b/jedi/parser_utils.py
@@ -293,6 +293,25 @@ def cut_value_at_position(leaf, position):
     return ''.join(lines)
 
 
+def expr_is_dotted(node):
+    """
+    Checks if a path looks like `name` or `name.foo.bar` and not `name()`.
+    """
+    if node.type == 'atom':
+        if len(node.children) == 3 and node.children[0] == '(':
+            return expr_is_dotted(node.children[1])
+        return False
+    if node.type == 'atom_expr':
+        children = node.children
+        if children[0] == 'await':
+            return False
+        if not expr_is_dotted(children[0]):
+            return False
+        # Check trailers
+        return all(c.children[0] == '.' for c in children[1:])
+    return node.type == 'name'
+
+
 def _function_is_x_method(method_name):
     def wrapper(function_node):
         """
diff --git a/test/completion/pep0484_typing.py b/test/completion/pep0484_typing.py
index 4ebab2e9..73ab32ee 100644
--- a/test/completion/pep0484_typing.py
+++ b/test/completion/pep0484_typing.py
@@ -499,3 +499,99 @@ def dynamic_annotation(x: int):
 
 #? int()
 dynamic_annotation('')
+
+# -------------------------
+# TypeDict
+# -------------------------
+
+# python >= 3.8
+
+class Foo(typing.TypedDict):
+    foo: str
+    bar: typing.List[float]
+    an_int: int
+    #! ['foo: str']
+    foo
+    #? str()
+    foo
+    #? int()
+    an_int
+
+def typed_dict_test_foo(arg: Foo):
+    a_string = arg['foo']
+    a_list_of_floats = arg['bar']
+    an_int = arg['an_int']
+
+    #? str()
+    a_string
+    #? list()
+    a_list_of_floats
+    #? float()
+    a_list_of_floats[0]
+    #? int()
+    an_int
+
+    #? ['isupper']
+    a_string.isuppe
+    #? ['pop']
+    a_list_of_floats.po
+    #? ['as_integer_ratio']
+    an_int.as_integer_rati
+
+#! ['class Foo']
+d: Foo
+#? str()
+d['foo']
+#? float()
+d['bar'][0]
+#?
+d['baz']
+
+#?
+d.foo
+#?
+d.bar
+#! []
+d.foo
+
+#? []
+Foo.set
+#? ['setdefault']
+d.setdefaul
+
+#? 5 ["'foo"]
+d['fo']
+#? 5 ['"bar"']
+d["bar"]
+
+class Bar(Foo):
+    another_variable: int
+
+    #? int()
+    another_variable
+    #?
+    an_int
+
+def typed_dict_test_foo(arg: Bar):
+    a_string = arg['foo']
+    a_list_of_floats = arg['bar']
+    an_int = arg['an_int']
+    another_variable = arg['another_variable']
+
+    #? str()
+    a_string
+    #? list()
+    a_list_of_floats
+    #? float()
+    a_list_of_floats[0]
+    #? int()
+    an_int
+    #? int()
+    another_variable
+
+    #? ['isupper']
+    a_string.isuppe
+    #? ['pop']
+    a_list_of_floats.po
+    #? ['as_integer_ratio']
+    an_int.as_integer_rati