Platform for stateful agents: AI with advanced memory that can learn and self-improve over time.
feat: launch async agent loop (#2658)
cthomas2bd937c5f721
diff --git a/letta/server/rest_api/routers/v1/agents.py b/letta/server/rest_api/routers/v1/agents.pyindex b84f7dd48..fe7e28788 100644--- a/letta/server/rest_api/routers/v1/agents.py+++ b/letta/server/rest_api/routers/v1/agents.py@@ -665,17 +665,14 @@ async def send_message("""actor = await server.user_manager.get_actor_or_default_async(actor_id=actor_id)request_start_timestamp_ns = get_utc_timestamp_ns()- user_eligible = actor.organization_id not in ["org-4a3af5dd-4c6a-48cb-ac13-3f73ecaaa4bf", "org-4ab3f6e8-9a44-4bee-aeb6-c681cbbc7bf6"]# TODO: This is redundant, remove soonagent = await server.agent_manager.get_agent_by_id_async(agent_id, actor, include_relationships=["multi_agent_group"])agent_eligible = agent.enable_sleeptime or agent.agent_type == AgentType.sleeptime_agent or not agent.multi_agent_group- experimental_header = request_obj.headers.get("X-EXPERIMENTAL") or "false"- feature_enabled = settings.use_experimental or experimental_header.lower() == "true"model_compatible = agent.llm_config.model_endpoint_type in ["anthropic", "openai", "together", "google_ai", "google_vertex"]- if user_eligible and agent_eligible and feature_enabled and model_compatible:+ if agent_eligible and model_compatible:if agent.enable_sleeptime and agent.agent_type != AgentType.voice_convo_agent:- experimental_agent = SleeptimeMultiAgentV2(+ agent_loop = SleeptimeMultiAgentV2(agent_id=agent_id,message_manager=server.message_manager,agent_manager=server.agent_manager,@@ -687,7 +684,7 @@ async def send_message(group=agent.multi_agent_group,)else:- experimental_agent = LettaAgent(+ agent_loop = LettaAgent(agent_id=agent_id,message_manager=server.message_manager,agent_manager=server.agent_manager,@@ -698,7 +695,7 @@ async def send_message(telemetry_manager=server.telemetry_manager if settings.llm_api_logging else NoopTelemetryManager(),)- result = await experimental_agent.step(+ result = await agent_loop.step(request.messages,max_steps=10,use_assistant_message=request.use_assistant_message,@@ -746,20 +743,17 @@ async def send_message_streaming("""request_start_timestamp_ns = get_utc_timestamp_ns()actor = await server.user_manager.get_actor_or_default_async(actor_id=actor_id)- user_eligible = actor.organization_id not in ["org-4a3af5dd-4c6a-48cb-ac13-3f73ecaaa4bf", "org-4ab3f6e8-9a44-4bee-aeb6-c681cbbc7bf6"]# TODO: This is redundant, remove soonagent = await server.agent_manager.get_agent_by_id_async(agent_id, actor, include_relationships=["multi_agent_group"])agent_eligible = agent.enable_sleeptime or agent.agent_type == AgentType.sleeptime_agent or not agent.multi_agent_group- experimental_header = request_obj.headers.get("X-EXPERIMENTAL") or "false"- feature_enabled = settings.use_experimental or experimental_header.lower() == "true"model_compatible = agent.llm_config.model_endpoint_type in ["anthropic", "openai", "together", "google_ai", "google_vertex"]model_compatible_token_streaming = agent.llm_config.model_endpoint_type in ["anthropic", "openai"]not_letta_endpoint = not ("inference.letta.com" in agent.llm_config.model_endpoint)request_start_timestamp_ns = get_utc_timestamp_ns()- if user_eligible and agent_eligible and feature_enabled and model_compatible:+ if agent_eligible and model_compatible:if agent.enable_sleeptime and agent.agent_type != AgentType.voice_convo_agent:- experimental_agent = SleeptimeMultiAgentV2(+ agent_loop = SleeptimeMultiAgentV2(agent_id=agent_id,message_manager=server.message_manager,agent_manager=server.agent_manager,@@ -773,7 +767,7 @@ async def send_message_streaming(group=agent.multi_agent_group,)else:- experimental_agent = LettaAgent(+ agent_loop = LettaAgent(agent_id=agent_id,message_manager=server.message_manager,agent_manager=server.agent_manager,@@ -787,7 +781,7 @@ async def send_message_streaming(if request.stream_tokens and model_compatible_token_streaming and not_letta_endpoint:result = StreamingResponseWithStatusCode(- experimental_agent.step_stream(+ agent_loop.step_stream(input_messages=request.messages,max_steps=10,use_assistant_message=request.use_assistant_message,@@ -797,7 +791,7 @@ async def send_message_streaming()else:result = StreamingResponseWithStatusCode(- experimental_agent.step_stream_no_tokens(+ agent_loop.step_stream_no_tokens(request.messages,max_steps=10,use_assistant_message=request.use_assistant_message,@@ -964,11 +958,9 @@ async def summarize_agent_conversation(# TODO: This is redundant, remove soonagent = await server.agent_manager.get_agent_by_id_async(agent_id, actor, include_relationships=["multi_agent_group"])agent_eligible = agent.enable_sleeptime or agent.agent_type == AgentType.sleeptime_agent or not agent.multi_agent_group- experimental_header = request_obj.headers.get("X-EXPERIMENTAL") or "false"- feature_enabled = settings.use_experimental or experimental_header.lower() == "true"model_compatible = agent.llm_config.model_endpoint_type in ["anthropic", "openai", "together", "google_ai", "google_vertex"]- if agent_eligible and feature_enabled and model_compatible:+ if agent_eligible and model_compatible:agent = LettaAgent(agent_id=agent_id,message_manager=server.message_manager,diff --git a/tests/integration_test_builtin_tools.py b/tests/integration_test_builtin_tools.pyindex dfbad8e4c..ebb09d03d 100644--- a/tests/integration_test_builtin_tools.py+++ b/tests/integration_test_builtin_tools.py@@ -13,7 +13,6 @@ from letta_client.types import ToolReturnMessagefrom letta.schemas.agent import AgentStatefrom letta.schemas.llm_config import LLMConfig-from letta.settings import settings# ------------------------------# Fixtures@@ -54,10 +53,7 @@ def server_url() -> str:else:raise RuntimeError(f"Could not reach {url} within {timeout_seconds}s")- temp = settings.use_experimental- settings.use_experimental = Trueyield url- settings.use_experimental = temp@pytest.fixture(scope="module")diff --git a/tests/integration_test_multi_agent.py b/tests/integration_test_multi_agent.pyindex 748f824b6..6e9cbf8c9 100644--- a/tests/integration_test_multi_agent.py+++ b/tests/integration_test_multi_agent.py@@ -14,7 +14,6 @@ from letta.schemas.letta_message import SystemMessage, ToolReturnMessagefrom letta.schemas.tool import Toolfrom letta.server.server import SyncServerfrom letta.services.agent_manager import AgentManager-from letta.settings import settingsfrom tests.helpers.utils import retry_until_successfrom tests.utils import wait_for_incoming_message@@ -53,10 +52,7 @@ def server_url() -> str:else:raise RuntimeError(f"Could not reach {url} within {timeout_seconds}s")- temp = settings.use_experimental- settings.use_experimental = Trueyield url- settings.use_experimental = temp@pytest.fixture(scope="module")