diff --git a/src/haiku/rag/research/base.py b/src/haiku/rag/research/base.py
index 8098fc7e..e547a50f 100644
--- a/src/haiku/rag/research/base.py
+++ b/src/haiku/rag/research/base.py
@@ -1,7 +1,7 @@
 from abc import ABC, abstractmethod
 from typing import Any, Generic, TypeVar
 
-from pydantic import BaseModel
+from pydantic import BaseModel, Field
 from pydantic_ai import Agent
 from pydantic_ai.models.openai import OpenAIChatModel
 from pydantic_ai.providers.ollama import OllamaProvider
@@ -86,7 +86,7 @@ class SearchResult(BaseModel):
     content: str
     score: float
     document_uri: str
-    metadata: dict[str, Any] = {}
+    metadata: dict[str, Any] = Field(default_factory=dict)
 
 
 class ResearchOutput(BaseModel):
diff --git a/src/haiku/rag/research/orchestrator.py b/src/haiku/rag/research/orchestrator.py
index 278acc05..2e43c8d7 100644
--- a/src/haiku/rag/research/orchestrator.py
+++ b/src/haiku/rag/research/orchestrator.py
@@ -56,16 +56,12 @@ class ResearchOrchestrator(BaseResearchAgent[ResearchPlan]):
 
     def _format_context_for_prompt(self, context: ResearchContext) -> str:
         """Format the research context as XML for inclusion in prompts."""
+
         context_data = {
             "original_question": context.original_question,
             "unanswered_questions": context.sub_questions,
             "qa_responses": [
-                {
-                    "question": qa["question"],
-                    "answer": qa["answer"][:500] + "..."
-                    if len(qa["answer"]) > 500
-                    else qa["answer"],
-                }
+                {"question": qa["question"], "answer": qa["answer"]}
                 for qa in context.qa_responses
             ],
             "insights": context.insights,
@@ -201,7 +197,7 @@ Evaluate the research progress for the original question and identify any remain
                 output = evaluation_result.output
                 if output.key_insights:
                     console.print("   [bold]Key insights:[/bold]")
-                    for insight in output.key_insights[:3]:
+                    for insight in output.key_insights:
                         console.print(f"   • {insight}")
                 console.print(
                     f"   Confidence: [yellow]{output.confidence_score:.1%}[/yellow]"
diff --git a/src/haiku/rag/research/prompts.py b/src/haiku/rag/research/prompts.py
index a4b36d91..05ba7be3 100644
--- a/src/haiku/rag/research/prompts.py
+++ b/src/haiku/rag/research/prompts.py
@@ -22,6 +22,12 @@ Your role is to:
 4. Base your answer strictly on the information found
 
 Use the search_and_answer tool to retrieve relevant documents and formulate your response.
+
+Important:
+- Always call the search_and_answer tool before drafting any answer.
+- Never answer without first using the tool at least once.
+- If no relevant information is found, state that clearly and avoid speculation.
+
 Be thorough and specific in your answers, citing relevant information from the sources."""
 
 EVALUATION_AGENT_PROMPT = """You are an analysis and evaluation specialist for research workflows.