· 4 min read
Market research involves multiple independent analyses that converge into a single picture. You need market sizing, customer segmentation, competitive mapping, and trend analysis — and none of these depend on each other during execution.
That makes it a textbook case for the Parallel Workers pattern: four agents working simultaneously on distinct research tracks, with their outputs synthesized into a comprehensive market view.
Here's how to build one from scratch.
Role: Calculate Total Addressable Market (TAM), Serviceable Addressable Market (SAM), and Serviceable Obtainable Market (SOM).
This agent takes your product description and target geography, then builds market sizing from multiple angles. It should estimate top-down (from industry reports and macro data) and bottom-up (from customer counts and pricing), then reconcile the two approaches.
Key prompt elements:
Role: Identify and profile distinct customer segments within the target market.
This agent maps out who buys in your market, how they differ, and what drives their purchasing decisions. It goes beyond demographics into behavioral segmentation — how different groups evaluate, buy, and use products in your category.
Key prompt elements:
Role: Map direct and indirect competitors, their positioning, strengths, and vulnerabilities.
This agent builds a structured view of the competitive environment. It should identify direct competitors (same solution, same market), adjacent competitors (different solution, same problem), and potential entrants (companies with the capability and incentive to enter).
Key prompt elements:
Role: Identify macro trends, technology shifts, and regulatory changes affecting the market.
This agent looks at the forces shaping the market's future. It covers technology trends (what's becoming possible), regulatory trends (what's becoming required or restricted), buyer behavior trends (how purchasing is changing), and market structure trends (consolidation, fragmentation, platform shifts).
Key prompt elements:
The quality of your market research output depends heavily on the context you provide. Before running the team, prepare:
Product description: What you're building or selling, in two to three specific sentences. Include the core value proposition and primary use case.
Target market definition: Geography, industry vertical, company size range, and any other boundaries. Be specific — "mid-market SaaS companies in the US with 50-500 employees" is dramatically better than "businesses."
Known context: Share what you already know. If you have existing customer data, competitive intelligence, or market assumptions, include them. The agents build on your context rather than starting from zero.
Specific questions: If there are particular strategic questions driving this research ("Should we expand to Europe?" or "Is the SMB segment worth pursuing?"), include them. They help the agents focus their analysis.
Four parallel agents produce four separate outputs. The synthesis step is where the real value emerges.
Look for convergence — when multiple agents point to the same conclusion from different angles, that signal is strong. If the Market Size Analyst identifies a growing segment and the Trend Analyst flags a technology shift enabling that growth, you have a high-conviction opportunity.
Look for contradictions — when agents disagree, that's where the interesting strategic questions live. If the Competitive Landscape Mapper shows a crowded market but the Market Size Analyst shows rapid growth, that tension reveals whether the market is big enough for new entrants.
Agent teams produce plausible analysis, not ground truth. Always validate against known data points:
The agent team gives you a structured starting point in minutes. Validation turns that starting point into something you can make decisions on.
The first run rarely produces a perfect market picture. Use the output to refine your input:
Two or three iterations typically produce research quality that rivals a week-long manual effort. The total time investment: under an hour.