The quality of Deep Research output depends heavily on how you frame your research question and refine the plan.

Prompt Structure for Research

A strong research prompt includes:

• Topic — Clear subject area
• Scope — How broad or narrow
• Perspective — What angle to cover
• Audience — Who will read the report
• Specifics — Any particular aspects to include or exclude

Example Prompts

Market Research: "Research the current state of the AI code editor market in 2026. Cover major players (Cursor, Trae, Copilot, Claude Code), market size estimates, pricing models, key differentiators, and trends. Focus on adoption in enterprise development teams."

Technology Evaluation: "Compare vector database options for a production RAG system handling 10M+ documents. Evaluate Pinecone, Weaviate, Qdrant, and Milvus across performance, cost, ease of deployment, scalability, and ecosystem support."

Academic Research: "Survey recent studies (2024-2026) on the effects of AI-assisted writing on student learning outcomes in higher education. Include methodology assessments and conflicting findings."

Refining the Research Plan

After submitting your query, Deep Research presents a plan:

• Review each subtopic for relevance
• Add missing areas: "Also include pricing comparison tables"
• Remove unnecessary sections: "Skip the historical background"
• Adjust depth: "Go deeper on the technical architecture comparison"
• Specify format: "Include a summary table comparing all options"

Multi-Step Research

For complex topics, break into sequential queries:

1. Broad overview first: "What are the main approaches to..."
2. Deep dive on findings: "Now research [specific approach] in detail..."
3. Comparative analysis: "Compare [option A] vs [option B] based on..."
4. Practical recommendations: "Based on our research, what would you recommend for..."