The AI Research Framework

Every useful analysis starts with a question sharp enough to answer. Before you open a single tool, define the boundaries of what you are investigating and why the answer matters to a specific decision.

• What you are analyzing and why it matters now
• Scope: what is in, what is out
• Time horizon and decision type (buy/hold/pass, allocation shift, risk assessment)

AI can extract and summarize at speed, but the quality of your research depends on the quality of your sources. Start with primary data, then use AI to accelerate extraction while maintaining a verification discipline.

• Primary sources: SEC filings, earnings transcripts, economic data releases
• AI-assisted extraction: what to ask the model, what to verify independently
• Cross-reference claims against at least two independent sources before treating them as established

Structure your analysis as a clear pipeline: inputs flow through analysis to produce outputs. Define what the AI handles and what stays with you. The best results come from combining AI speed with human judgment, not from delegating the entire process.

• Structured framework: inputs, analysis steps, expected outputs
• AI role: summarization, pattern matching, scenario modeling
• Human role: judgment, context, sanity checks, domain expertise

Every claim needs a trail. Document where each piece of evidence came from, what assumptions you are making, and what the AI contributed versus what you added. This audit trail is what separates research from speculation.

• Source citation trail for every claim and data point
• Assumption log with confidence levels (high, medium, low)
• Clear labeling of what the AI contributed versus what you added or modified

Research without a decision framework is just reading. Translate your analysis into a clear proceed, hold, or pass decision. Document the risks, set monitoring triggers, and define what new information would change your conclusion.

• Decision matrix: proceed, hold, or pass with explicit criteria for each
• Risk factors and monitoring triggers that would prompt reassessment
• Pre-commitment: what specific evidence would change your conclusion