diff --git a/src/google/adk/memory/vertex_ai_memory_bank_service.py b/src/google/adk/memory/vertex_ai_memory_bank_service.py
index e52ef99d..af949fb2 100644
--- a/src/google/adk/memory/vertex_ai_memory_bank_service.py
+++ b/src/google/adk/memory/vertex_ai_memory_bank_service.py
@@ -33,6 +33,7 @@ from .memory_entry import MemoryEntry
 
 if TYPE_CHECKING:
   import vertexai
+  from vertexai import types as vertex_types
 
   from ..events.event import Event
   from ..sessions.session import Session
@@ -107,22 +108,21 @@ _MAX_DIRECT_MEMORIES_PER_GENERATE_CALL = 5
 def _supports_generate_memories_metadata() -> bool:
   """Returns whether installed Vertex SDK supports config.metadata."""
   try:
-    from vertexai._genai.types import common as vertex_common_types
+    from vertexai import types as vertex_types
   except ImportError:
     return False
   return (
-      'metadata'
-      in vertex_common_types.GenerateAgentEngineMemoriesConfig.model_fields
+      'metadata' in vertex_types.GenerateAgentEngineMemoriesConfig.model_fields
   )
 
 
 def _supports_create_memory_metadata() -> bool:
   """Returns whether installed Vertex SDK supports create config.metadata."""
   try:
-    from vertexai._genai.types import common as vertex_common_types
+    from vertexai import types as vertex_types
   except ImportError:
     return False
-  return 'metadata' in vertex_common_types.AgentEngineMemoryConfig.model_fields
+  return 'metadata' in vertex_types.AgentEngineMemoryConfig.model_fields
 
 
 @lru_cache(maxsize=1)
@@ -133,14 +133,12 @@ def _get_generate_memories_config_keys() -> frozenset[str]:
   allowlist to preserve compatibility when introspection is unavailable.
   """
   try:
-    from vertexai._genai.types import common as vertex_common_types
+    from vertexai import types as vertex_types
   except ImportError:
     return _GENERATE_MEMORIES_CONFIG_FALLBACK_KEYS
 
   try:
-    model_fields = (
-        vertex_common_types.GenerateAgentEngineMemoriesConfig.model_fields
-    )
+    model_fields = vertex_types.GenerateAgentEngineMemoriesConfig.model_fields
   except AttributeError:
     return _GENERATE_MEMORIES_CONFIG_FALLBACK_KEYS
 
@@ -157,12 +155,12 @@ def _get_create_memory_config_keys() -> frozenset[str]:
   allowlist to preserve compatibility when introspection is unavailable.
   """
   try:
-    from vertexai._genai.types import common as vertex_common_types
+    from vertexai import types as vertex_types
   except ImportError:
     return _CREATE_MEMORY_CONFIG_FALLBACK_KEYS
 
   try:
-    model_fields = vertex_common_types.AgentEngineMemoryConfig.model_fields
+    model_fields = vertex_types.AgentEngineMemoryConfig.model_fields
   except AttributeError:
     return _CREATE_MEMORY_CONFIG_FALLBACK_KEYS
 
@@ -574,6 +572,39 @@ class VertexAiMemoryBankService(BaseMemoryService):
       )
     return SearchMemoryResponse(memories=memory_events)
 
+  async def retrieve_profiles(
+      self,
+      *,
+      app_name: str,
+      user_id: str,
+  ) -> list[vertex_types.MemoryProfile]:
+    """Retrieves structured user profiles for the scope, one per schema.
+
+    Profiles are a Vertex Memory Bank capability distinct from memory search:
+    a scope-keyed lookup, not a semantic query.
+
+    Args:
+      app_name: The application name for the profile scope.
+      user_id: The user ID for the profile scope.
+
+    Returns:
+      The structured profiles for the scope, one per registered schema.
+    """
+    api_client = self._get_api_client()
+    response = await api_client.agent_engines.memories.retrieve_profiles(
+        name='reasoningEngines/' + self._agent_engine_id,
+        scope={
+            'app_name': app_name,
+            'user_id': user_id,
+        },
+    )
+    profiles = list((response.profiles or {}).values())
+    if profiles:
+      logger.info('Retrieved %d memory profiles.', len(profiles))
+    else:
+      logger.info('Retrieved no memory profiles.')
+    return profiles
+
   def _get_api_client(self) -> vertexai.AsyncClient:
     """Instantiates an API client for the given project and location.
 
diff --git a/src/google/adk/tools/__init__.py b/src/google/adk/tools/__init__.py
index 770a2cb3..641c5c20 100644
--- a/src/google/adk/tools/__init__.py
+++ b/src/google/adk/tools/__init__.py
@@ -42,6 +42,7 @@ if TYPE_CHECKING:
   from .transfer_to_agent_tool import transfer_to_agent
   from .transfer_to_agent_tool import TransferToAgentTool
   from .url_context_tool import url_context
+  from .vertex_ai_load_profiles_tool import VertexAiLoadProfilesTool
   from .vertex_ai_search_tool import VertexAiSearchTool
 
 # If you are adding a new tool to this file, please make sure you add it to the
@@ -89,6 +90,10 @@ _LAZY_MAPPING = {
         'TransferToAgentTool',
     ),
     'url_context': ('.url_context_tool', 'url_context'),
+    'VertexAiLoadProfilesTool': (
+        '.vertex_ai_load_profiles_tool',
+        'VertexAiLoadProfilesTool',
+    ),
     'VertexAiSearchTool': ('.vertex_ai_search_tool', 'VertexAiSearchTool'),
     'MCPToolset': ('.mcp_tool.mcp_toolset', 'MCPToolset'),
     'McpToolset': ('.mcp_tool.mcp_toolset', 'McpToolset'),
diff --git a/src/google/adk/tools/vertex_ai_load_profiles_tool.py b/src/google/adk/tools/vertex_ai_load_profiles_tool.py
new file mode 100644
index 00000000..26d8b87b
--- /dev/null
+++ b/src/google/adk/tools/vertex_ai_load_profiles_tool.py
@@ -0,0 +1,67 @@
+# Copyright 2026 Google LLC
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+from __future__ import annotations
+
+from typing import Any
+from typing import TYPE_CHECKING
+
+from google.genai import types
+from typing_extensions import override
+
+from ..features import FeatureName
+from ..features import is_feature_enabled
+from .function_tool import FunctionTool
+from .tool_context import ToolContext
+
+if TYPE_CHECKING:
+  from ..memory.vertex_ai_memory_bank_service import VertexAiMemoryBankService
+
+
+class VertexAiLoadProfilesTool(FunctionTool):
+  """A tool that loads a user's structured profiles from Vertex Memory Bank."""
+
+  def __init__(self, memory_service: VertexAiMemoryBankService):
+    super().__init__(self.load_profiles)
+    self._memory_service = memory_service
+
+  async def load_profiles(self, tool_context: ToolContext) -> dict[str, Any]:
+    """Loads structured user profiles for the current user."""
+    profiles = await self._memory_service.retrieve_profiles(
+        app_name=tool_context.session.app_name,
+        user_id=tool_context.user_id,
+    )
+    return {
+        'profiles': [profile.profile for profile in profiles if profile.profile]
+    }
+
+  @override
+  def _get_declaration(self) -> types.FunctionDeclaration | None:
+    if is_feature_enabled(FeatureName.JSON_SCHEMA_FOR_FUNC_DECL):
+      return types.FunctionDeclaration(
+          name=self.name,
+          description=self.description,
+          parameters_json_schema={
+              'type': 'object',
+              'properties': {},
+          },
+      )
+    return types.FunctionDeclaration(
+        name=self.name,
+        description=self.description,
+        parameters=types.Schema(
+            type=types.Type.OBJECT,
+            properties={},
+        ),
+    )
diff --git a/tests/unittests/memory/test_vertex_ai_memory_bank_service.py b/tests/unittests/memory/test_vertex_ai_memory_bank_service.py
index e41e797f..529c3aec 100644
--- a/tests/unittests/memory/test_vertex_ai_memory_bank_service.py
+++ b/tests/unittests/memory/test_vertex_ai_memory_bank_service.py
@@ -14,6 +14,7 @@
 
 import asyncio
 import datetime
+import logging
 from typing import Any
 from typing import Iterable
 from typing import Optional
@@ -26,7 +27,7 @@ from google.adk.memory.vertex_ai_memory_bank_service import VertexAiMemoryBankSe
 from google.adk.sessions.session import Session
 from google.genai import types
 import pytest
-from vertexai._genai.types import common as vertex_common_types
+from vertexai import types as vertex_types
 
 MOCK_APP_NAME = 'test-app'
 MOCK_USER_ID = 'test-user'
@@ -34,20 +35,16 @@ MOCK_USER_ID = 'test-user'
 
 def _supports_generate_memories_metadata() -> bool:
   return (
-      'metadata'
-      in vertex_common_types.GenerateAgentEngineMemoriesConfig.model_fields
+      'metadata' in vertex_types.GenerateAgentEngineMemoriesConfig.model_fields
   )
 
 
 def _supports_create_memory_metadata() -> bool:
-  return 'metadata' in vertex_common_types.AgentEngineMemoryConfig.model_fields
+  return 'metadata' in vertex_types.AgentEngineMemoryConfig.model_fields
 
 
 def _supports_create_memory_revision_labels() -> bool:
-  return (
-      'revision_labels'
-      in vertex_common_types.AgentEngineMemoryConfig.model_fields
-  )
+  return 'revision_labels' in vertex_types.AgentEngineMemoryConfig.model_fields
 
 
 class _AsyncListIterator:
@@ -208,6 +205,9 @@ def mock_vertexai_client():
     mock_async_client.agent_engines.memories.generate = mock.AsyncMock()
     mock_async_client.agent_engines.memories.create = mock.AsyncMock()
     mock_async_client.agent_engines.memories.retrieve = mock.AsyncMock()
+    mock_async_client.agent_engines.memories.retrieve_profiles = (
+        mock.AsyncMock()
+    )
     mock_async_client.agent_engines.memories.ingest_events = mock.AsyncMock()
 
     mock_client = mock.MagicMock()
@@ -305,7 +305,7 @@ async def test_add_events_to_memory_with_explicit_events_and_metadata(
   source = call_kwargs['direct_contents_source']
   assert len(source.events) == 1
   assert source.events[0].content.parts[0].text == 'test_content'
-  vertex_common_types.GenerateAgentEngineMemoriesConfig(**call_kwargs['config'])
+  vertex_types.GenerateAgentEngineMemoriesConfig(**call_kwargs['config'])
 
 
 @pytest.mark.asyncio
@@ -336,7 +336,7 @@ async def test_add_events_to_memory_without_session_id(
   source = call_kwargs['direct_contents_source']
   assert len(source.events) == 1
   assert source.events[0].content.parts[0].text == 'test_content'
-  vertex_common_types.GenerateAgentEngineMemoriesConfig(**call_kwargs['config'])
+  vertex_types.GenerateAgentEngineMemoriesConfig(**call_kwargs['config'])
   mock_vertexai_client.agent_engines.memories.create.assert_not_called()
 
 
@@ -376,7 +376,7 @@ async def test_add_events_to_memory_merges_metadata_field_and_unknown_keys(
   source = call_kwargs['direct_contents_source']
   assert len(source.events) == 1
   assert source.events[0].content.parts[0].text == 'test_content'
-  vertex_common_types.GenerateAgentEngineMemoriesConfig(**call_kwargs['config'])
+  vertex_types.GenerateAgentEngineMemoriesConfig(**call_kwargs['config'])
 
 
 @pytest.mark.asyncio
@@ -407,7 +407,7 @@ async def test_add_events_to_memory_none_wait_for_completion_keeps_default(
   source = call_kwargs['direct_contents_source']
   assert len(source.events) == 1
   assert source.events[0].content.parts[0].text == 'test_content'
-  vertex_common_types.GenerateAgentEngineMemoriesConfig(**call_kwargs['config'])
+  vertex_types.GenerateAgentEngineMemoriesConfig(**call_kwargs['config'])
 
 
 @pytest.mark.asyncio
@@ -442,7 +442,7 @@ async def test_add_events_to_memory_ttl_used_when_revision_ttl_is_none(
   source = call_kwargs['direct_contents_source']
   assert len(source.events) == 1
   assert source.events[0].content.parts[0].text == 'test_content'
-  vertex_common_types.GenerateAgentEngineMemoriesConfig(**call_kwargs['config'])
+  vertex_types.GenerateAgentEngineMemoriesConfig(**call_kwargs['config'])
 
 
 @pytest.mark.asyncio
@@ -587,7 +587,7 @@ async def test_add_memory_calls_create(
           'config'
       ]
   )
-  vertex_common_types.AgentEngineMemoryConfig(**create_config)
+  vertex_types.AgentEngineMemoryConfig(**create_config)
 
 
 @pytest.mark.asyncio
@@ -634,7 +634,7 @@ async def test_add_memory_enable_consolidation_calls_generate_direct_source(
           'config'
       ]
   )
-  vertex_common_types.GenerateAgentEngineMemoriesConfig(**generate_config)
+  vertex_types.GenerateAgentEngineMemoriesConfig(**generate_config)
 
 
 @pytest.mark.asyncio
@@ -768,7 +768,7 @@ async def test_add_memory_calls_create_with_memory_entry_metadata(
           'config'
       ]
   )
-  vertex_common_types.AgentEngineMemoryConfig(**create_config)
+  vertex_types.AgentEngineMemoryConfig(**create_config)
 
 
 @pytest.mark.asyncio
@@ -1009,6 +1009,66 @@ async def test_search_memory_empty_results(mock_vertexai_client):
   assert len(result.memories) == 0
 
 
+@pytest.mark.asyncio
+async def test_retrieve_profiles(mock_vertexai_client, caplog):
+  """Returns the structured profiles for the scope as a list."""
+  retrieve_profiles_response = vertex_types.RetrieveProfilesResponse(
+      profiles={
+          'user-profile': vertex_types.MemoryProfile(
+              schema_id='user-profile',
+              profile={'name': 'Kim'},
+          )
+      }
+  )
+  mock_vertexai_client.agent_engines.memories.retrieve_profiles.return_value = (
+      retrieve_profiles_response
+  )
+  memory_service = mock_vertex_ai_memory_bank_service()
+
+  with caplog.at_level(logging.INFO):
+    result = await memory_service.retrieve_profiles(
+        app_name=MOCK_APP_NAME,
+        user_id=MOCK_USER_ID,
+    )
+
+  mock_vertexai_client.agent_engines.memories.retrieve_profiles.assert_awaited_once_with(
+      name='reasoningEngines/123',
+      scope={'app_name': MOCK_APP_NAME, 'user_id': MOCK_USER_ID},
+  )
+  assert 'Retrieved 1 memory profiles.' in caplog.text
+  assert result == [
+      vertex_types.MemoryProfile(
+          schema_id='user-profile',
+          profile={'name': 'Kim'},
+      )
+  ]
+
+
+@pytest.mark.asyncio
+async def test_retrieve_profiles_empty_results(mock_vertexai_client, caplog):
+  """Returns an empty list when the scope has no profiles."""
+  retrieve_profiles_response = vertex_types.RetrieveProfilesResponse(
+      profiles=None
+  )
+  mock_vertexai_client.agent_engines.memories.retrieve_profiles.return_value = (
+      retrieve_profiles_response
+  )
+  memory_service = mock_vertex_ai_memory_bank_service()
+
+  with caplog.at_level(logging.INFO):
+    result = await memory_service.retrieve_profiles(
+        app_name=MOCK_APP_NAME,
+        user_id=MOCK_USER_ID,
+    )
+
+  mock_vertexai_client.agent_engines.memories.retrieve_profiles.assert_awaited_once_with(
+      name='reasoningEngines/123',
+      scope={'app_name': MOCK_APP_NAME, 'user_id': MOCK_USER_ID},
+  )
+  assert 'Retrieved no memory profiles.' in caplog.text
+  assert not result
+
+
 async def test_search_memory_uses_async_client_path():
   sync_client = mock.MagicMock()
   sync_client.agent_engines.memories.retrieve.side_effect = AssertionError(
diff --git a/tests/unittests/tools/test_vertex_ai_load_profiles_tool.py b/tests/unittests/tools/test_vertex_ai_load_profiles_tool.py
new file mode 100644
index 00000000..81e4e308
--- /dev/null
+++ b/tests/unittests/tools/test_vertex_ai_load_profiles_tool.py
@@ -0,0 +1,100 @@
+# Copyright 2026 Google LLC
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+from types import SimpleNamespace
+
+from google.adk.features import FeatureName
+from google.adk.features._feature_registry import temporary_feature_override
+from google.adk.models.llm_request import LlmRequest
+from google.adk.tools.vertex_ai_load_profiles_tool import VertexAiLoadProfilesTool
+from pytest import mark
+from vertexai import types as vertex_types
+
+
+class _FakeMemoryService:
+  """Minimal profile-providing service for VertexAiLoadProfilesTool tests."""
+
+  def __init__(self, profiles):
+    self._profiles = profiles
+    self.calls = []
+
+  async def retrieve_profiles(self, *, app_name, user_id):
+    self.calls.append((app_name, user_id))
+    return self._profiles
+
+
+class _StubToolContext:
+  """Minimal ToolContext stub exposing only the scope the tool reads."""
+
+  def __init__(self, *, app_name='test-app', user_id='test-user'):
+    self.session = SimpleNamespace(app_name=app_name)
+    self.user_id = user_id
+
+
+@mark.asyncio
+async def test_load_profiles_returns_profile_payloads():
+  memory_service = _FakeMemoryService([
+      vertex_types.MemoryProfile(
+          schema_id='user-profile', profile={'name': 'Kim'}
+      ),
+      vertex_types.MemoryProfile(schema_id='empty', profile={}),
+  ])
+  tool = VertexAiLoadProfilesTool(memory_service=memory_service)
+
+  result = await tool.load_profiles(_StubToolContext())
+
+  assert result == {'profiles': [{'name': 'Kim'}]}
+  assert memory_service.calls == [('test-app', 'test-user')]
+
+
+def test_get_declaration_with_json_schema_feature_disabled():
+  tool = VertexAiLoadProfilesTool(memory_service=_FakeMemoryService([]))
+  with temporary_feature_override(FeatureName.JSON_SCHEMA_FOR_FUNC_DECL, False):
+    declaration = tool._get_declaration()
+
+  assert declaration.name == 'load_profiles'
+  assert declaration.parameters_json_schema is None
+  assert declaration.parameters.properties == {}
+
+
+def test_get_declaration_with_json_schema_feature_enabled():
+  tool = VertexAiLoadProfilesTool(memory_service=_FakeMemoryService([]))
+  with temporary_feature_override(FeatureName.JSON_SCHEMA_FOR_FUNC_DECL, True):
+    declaration = tool._get_declaration()
+
+  assert declaration.name == 'load_profiles'
+  assert declaration.parameters is None
+  assert declaration.parameters_json_schema == {
+      'type': 'object',
+      'properties': {},
+  }
+
+
+@mark.asyncio
+async def test_process_llm_request_registers_tool_only():
+  tool = VertexAiLoadProfilesTool(memory_service=_FakeMemoryService([]))
+  llm_request = LlmRequest()
+
+  await tool.process_llm_request(
+      tool_context=_StubToolContext(),
+      llm_request=llm_request,
+  )
+
+  assert llm_request.config.system_instruction is None
+  assert llm_request.config.tools is not None
+  assert llm_request.config.tools[0].function_declarations is not None
+  assert llm_request.config.tools[0].function_declarations[0].name == (
+      'load_profiles'
+  )
+  assert 'load_profiles' in llm_request.tools_dict