Autonomous Research Pipeline

# Autonomous Research Pipeline

You are an AI research systems architect. Build a fully autonomous research pipeline where an AI agent formulates its own sub-questions, gathers multi-source evidence, cross-validates findings, and produces a rigorous research report.

## Research Mission

**Central research question:** [RESEARCH_QUESTION]
**Depth level:** [DEPTH — Academic rigor / Executive brief / Rapid scan]
**Domain:** [DOMAIN — e.g., market research, scientific literature, competitive intelligence]
**Output audience:** [AUDIENCE]

## Pipeline Stages

### Stage 1: Question Decomposition
The agent must decompose [RESEARCH_QUESTION] into [SUB_QUESTION_COUNT] specific, answerable sub-questions. For each sub-question, identify: the ideal source type, the key search terms, and what a satisfying answer looks like.

### Stage 2: Parallel Evidence Gathering
For each sub-question, the agent runs [SEARCHES_PER_QUESTION] independent searches across [SOURCE_TYPES]. It must collect a minimum of [MIN_SOURCES_PER_QUESTION] unique sources per sub-question.

### Stage 3: Cross-Validation
For each finding, the agent must find at least [CONFIRMATION_COUNT] independent sources that corroborate it. Findings with fewer sources are flagged as [CONFIDENCE_LOW].

### Stage 4: Synthesis and Gap Analysis
Synthesize findings across all sub-questions. Identify:
- Where evidence converges (high-confidence conclusions)
- Where evidence conflicts (contested claims — present both sides)
- Where evidence is absent (gaps that remain unanswered)

### Stage 5: Report Generation
Produce a [FORMAT] report with: executive summary, methodology note, findings by sub-question, confidence levels for each conclusion, and a research limitations section.