Competitive analysis and market research with web search. Produces structured briefs from live data.
OVERVIEW
The research pipeline takes a topic or competitive landscape and produces structured intelligence briefs from live web data. Three phases scan the market, analyze competitors, and synthesize findings into actionable recommendations. Each phase writes to an output directory, and the next phase reads from disk -- no context window limitations, no lost information.
ARCHITECTURE
Web search across industry sources. Gathers raw data from competitor sites, documentation, pricing pages, and industry reports. Produces a structured raw findings document.
Reads the market scan output and produces a structured comparison matrix and architecture analysis. Feature-by-feature breakdown across all identified competitors.
Reads both previous phases and produces an executive brief with positioning recommendations. Identifies gaps, opportunities, and strategic differentiators.
DESIGN
Agents access live data, not training cutoff. Research reflects the current state of the market, not what the model last saw during training.
Each phase writes to output_dir, next phase reads from disk. No context window truncation. Full fidelity between phases.
Phase ordering is enforced through explicit depends_on declarations. No race conditions, no out-of-order execution.
RESULTS