Research
Research Maps: Decomposing a Topic into 100+ Sub-Niches
March 16, 2026 · 4 min read
A research map is the blueprint behind every Clerk investigation — 100+ sub-niches, each with timeframes, search parameters, and a unique ID.
You cannot deeply research "Web3" as one query. You research "layer-2 scaling", "zero-knowledge proofs", "DeFi lending", and ninety-seven other sub-niches — then you combine them.
That is the idea behind the Research Map, the output of Clerk’s Decomposition phase.
A research map is a JSON document containing 100+ sub-niches. Each sub-niche has a unique ID, a topic, a description, a timeframe (for example 2014–2026), and search parameters directing the workers to the right sources.
The map also stores the overall timeframe, an estimated duration, and a research ID like "research-web3-2026".
The power of the map is parallelization. Instead of one model skimming a topic, 100 focused investigations run simultaneously — each going deeper than a single pass ever could.
When combined with reliability scoring and statistical aggregation, this is what moves research from AI guessing to data science.