diff --git a/docs/agents.md b/docs/agents.md
index 9b310476..44b2d1a0 100644
--- a/docs/agents.md
+++ b/docs/agents.md
@@ -59,7 +59,7 @@ stateDiagram-v2
 Key nodes:
 
 - **Plan**: Decomposes the question into focused sub-questions
-- **Search (parallel)**: Answers all sub-questions in parallel
+- **Search (parallel)**: Answers sub-questions in parallel (respects max_concurrency)
 - **Decision**: Evaluates if we have sufficient information or need another iteration
 - **Synthesize**: Generates the final comprehensive answer
 
@@ -69,7 +69,7 @@ Key differences from Research:
 - **Direct answers**: Returns just the answer (not a full research report)
 - **Question-focused**: Optimized for answering specific questions, not open-ended research
 - **Supports citations**: Can include inline source citations like `[document.md]`
-- **Configurable iterations**: Control max_iterations (default: 2)
+- **Configurable iterations**: Control max_iterations (default: 2) and max_concurrency (default: 1)
 
 CLI usage:
 
@@ -99,7 +99,8 @@ async with HaikuRAG(path_to_db) as client:
     state = DeepQAState(
         context=context,
         max_sub_questions=3,
-        max_iterations=2
+        max_iterations=2,
+        max_concurrency=1
     )
     deps = DeepQADeps(client=client)
 
@@ -175,6 +176,7 @@ async with HaikuRAG(path_to_db) as client:
         context=ResearchContext(original_question=question),
         max_iterations=2,
         confidence_threshold=0.8,
+        max_concurrency=2,
     )
     deps = ResearchDeps(client=client)
 
@@ -209,6 +211,7 @@ async with HaikuRAG(path_to_db) as client:
         context=ResearchContext(original_question=question),
         max_iterations=2,
         confidence_threshold=0.8,
+        max_concurrency=2,
     )
     deps = ResearchDeps(client=client)
 
diff --git a/haiku_rag_slim/haiku/rag/mcp.py b/haiku_rag_slim/haiku/rag/mcp.py
index 612022ae..8a4573fa 100644
--- a/haiku_rag_slim/haiku/rag/mcp.py
+++ b/haiku_rag_slim/haiku/rag/mcp.py
@@ -2,11 +2,11 @@ from pathlib import Path
 from typing import Any
 
 from fastmcp import FastMCP
+from haiku.rag.client import HaikuRAG
+from haiku.rag.research.models import ResearchReport
 from pydantic import BaseModel
 
-from haiku.rag.client import HaikuRAG
 from haiku.rag.config import Config
-from haiku.rag.research.models import ResearchReport
 
 
 class SearchResult(BaseModel):
@@ -191,11 +191,12 @@ def create_mcp_server(db_path: Path) -> FastMCP:
         try:
             async with HaikuRAG(db_path) as rag:
                 if deep:
-                    from haiku.rag.config import Config
                     from haiku.rag.qa.deep.dependencies import DeepQAContext
                     from haiku.rag.qa.deep.graph import build_deep_qa_graph
                     from haiku.rag.qa.deep.state import DeepQADeps, DeepQAState
 
+                    from haiku.rag.config import Config
+
                     graph = build_deep_qa_graph(
                         provider=Config.qa.provider,
                         model=Config.qa.model,
@@ -219,6 +220,7 @@ def create_mcp_server(db_path: Path) -> FastMCP:
         question: str,
         max_iterations: int = 3,
         confidence_threshold: float = 0.8,
+        max_concurrency: int = 1,
     ) -> ResearchReport | None:
         """Run multi-agent research to investigate a complex question.
 
@@ -229,6 +231,7 @@ def create_mcp_server(db_path: Path) -> FastMCP:
             question: The research question to investigate.
             max_iterations: Maximum search/analyze iterations (default: 3).
             confidence_threshold: Minimum confidence score (0-1) to stop early (default: 0.8).
+            max_concurrency: Maximum concurrent sub-questions to process (default: 1).
 
         Returns:
             A research report with findings, or None if an error occurred.
@@ -247,6 +250,7 @@ def create_mcp_server(db_path: Path) -> FastMCP:
                     context=ResearchContext(original_question=question),
                     max_iterations=max_iterations,
                     confidence_threshold=confidence_threshold,
+                    max_concurrency=max_concurrency,
                 )
                 deps = ResearchDeps(client=rag)
 
diff --git a/haiku_rag_slim/haiku/rag/qa/deep/graph.py b/haiku_rag_slim/haiku/rag/qa/deep/graph.py
index d1e28fe2..4be28cd4 100644
--- a/haiku_rag_slim/haiku/rag/qa/deep/graph.py
+++ b/haiku_rag_slim/haiku/rag/qa/deep/graph.py
@@ -1,11 +1,5 @@
 from typing import Any
 
-from pydantic_ai import Agent, RunContext
-from pydantic_ai.format_prompt import format_as_xml
-from pydantic_ai.output import ToolOutput
-from pydantic_graph.beta import Graph, GraphBuilder, StepContext
-from pydantic_graph.beta.join import reduce_list_append
-
 from haiku.rag.graph_common import get_model, log
 from haiku.rag.graph_common.models import ResearchPlan, SearchAnswer
 from haiku.rag.graph_common.prompts import PLAN_PROMPT, SEARCH_AGENT_PROMPT
@@ -17,6 +11,11 @@ from haiku.rag.qa.deep.prompts import (
     SYNTHESIS_PROMPT_WITH_CITATIONS,
 )
 from haiku.rag.qa.deep.state import DeepQADeps, DeepQAState
+from pydantic_ai import Agent, RunContext
+from pydantic_ai.format_prompt import format_as_xml
+from pydantic_ai.output import ToolOutput
+from pydantic_graph.beta import Graph, GraphBuilder, StepContext
+from pydantic_graph.beta.join import reduce_list_append
 
 
 def build_deep_qa_graph(
@@ -85,6 +84,21 @@ def build_deep_qa_graph(
         deps = ctx.deps
         sub_q = ctx.inputs
 
+        # Create semaphore if not already provided
+        if deps.semaphore is None:
+            import asyncio
+
+            deps.semaphore = asyncio.Semaphore(state.max_concurrency)
+
+        # Use semaphore to control concurrency
+        async with deps.semaphore:
+            return await _do_search(state, deps, sub_q)
+
+    async def _do_search(
+        state: DeepQAState,
+        deps: DeepQADeps,
+        sub_q: str,
+    ) -> SearchAnswer:
         log(
             deps,
             state,
diff --git a/haiku_rag_slim/haiku/rag/qa/deep/state.py b/haiku_rag_slim/haiku/rag/qa/deep/state.py
index 95c5bd5f..46750242 100644
--- a/haiku_rag_slim/haiku/rag/qa/deep/state.py
+++ b/haiku_rag_slim/haiku/rag/qa/deep/state.py
@@ -1,15 +1,16 @@
+import asyncio
 from dataclasses import dataclass
 
-from rich.console import Console
-
 from haiku.rag.client import HaikuRAG
 from haiku.rag.qa.deep.dependencies import DeepQAContext
+from rich.console import Console
 
 
 @dataclass
 class DeepQADeps:
     client: HaikuRAG
     console: Console | None = None
+    semaphore: asyncio.Semaphore | None = None
 
     def emit_log(self, message: str, state: "DeepQAState | None" = None) -> None:
         if self.console:
@@ -21,4 +22,5 @@ class DeepQAState:
     context: DeepQAContext
     max_sub_questions: int = 3
     max_iterations: int = 2
+    max_concurrency: int = 1
     iterations: int = 0
diff --git a/haiku_rag_slim/haiku/rag/research/graph.py b/haiku_rag_slim/haiku/rag/research/graph.py
index 22914bbf..117bf741 100644
--- a/haiku_rag_slim/haiku/rag/research/graph.py
+++ b/haiku_rag_slim/haiku/rag/research/graph.py
@@ -1,11 +1,5 @@
 from typing import Any
 
-from pydantic_ai import Agent, RunContext
-from pydantic_ai.format_prompt import format_as_xml
-from pydantic_ai.output import ToolOutput
-from pydantic_graph.beta import Graph, GraphBuilder, StepContext
-from pydantic_graph.beta.join import reduce_list_append
-
 from haiku.rag.graph_common import get_model, log
 from haiku.rag.graph_common.models import ResearchPlan, SearchAnswer
 from haiku.rag.graph_common.prompts import PLAN_PROMPT, SEARCH_AGENT_PROMPT
@@ -25,6 +19,11 @@ from haiku.rag.research.prompts import (
     SYNTHESIS_AGENT_PROMPT,
 )
 from haiku.rag.research.state import ResearchDeps, ResearchState
+from pydantic_ai import Agent, RunContext
+from pydantic_ai.format_prompt import format_as_xml
+from pydantic_ai.output import ToolOutput
+from pydantic_graph.beta import Graph, GraphBuilder, StepContext
+from pydantic_graph.beta.join import reduce_list_append
 
 
 def build_research_graph(
@@ -94,6 +93,21 @@ def build_research_graph(
         deps = ctx.deps
         sub_q = ctx.inputs
 
+        # Create semaphore if not already provided
+        if deps.semaphore is None:
+            import asyncio
+
+            deps.semaphore = asyncio.Semaphore(state.max_concurrency)
+
+        # Use semaphore to control concurrency
+        async with deps.semaphore:
+            return await _do_search(state, deps, sub_q)
+
+    async def _do_search(
+        state: ResearchState,
+        deps: ResearchDeps,
+        sub_q: str,
+    ) -> SearchAnswer:
         log(
             deps,
             state,
diff --git a/haiku_rag_slim/haiku/rag/research/state.py b/haiku_rag_slim/haiku/rag/research/state.py
index 989687f9..2c748103 100644
--- a/haiku_rag_slim/haiku/rag/research/state.py
+++ b/haiku_rag_slim/haiku/rag/research/state.py
@@ -1,11 +1,11 @@
+import asyncio
 from dataclasses import dataclass
 
-from rich.console import Console
-
 from haiku.rag.client import HaikuRAG
 from haiku.rag.research.dependencies import ResearchContext
 from haiku.rag.research.models import EvaluationResult, InsightAnalysis
 from haiku.rag.research.stream import ResearchStream
+from rich.console import Console
 
 
 @dataclass
@@ -13,6 +13,7 @@ class ResearchDeps:
     client: HaikuRAG
     console: Console | None = None
     stream: ResearchStream | None = None
+    semaphore: asyncio.Semaphore | None = None
 
     def emit_log(self, message: str, state: "ResearchState | None" = None) -> None:
         if self.console:
@@ -27,5 +28,6 @@ class ResearchState:
     iterations: int = 0
     max_iterations: int = 3
     confidence_threshold: float = 0.8
+    max_concurrency: int = 1
     last_eval: EvaluationResult | None = None
     last_analysis: InsightAnalysis | None = None