02 / CASE STUDY2026 / AI research automation
Prospect Research Agent
PROSPECT RESEARCH AGENT
JOB / PR-041RUNNING
01DISCOVER1,114 URLs
02VERIFYVERIFY DOMAIN
03SCOREICP SCORE
04BUYER361 BUYERS
05CHECK331 HUMAN
42.8×BEFORE
4.12×AFTER
1,557TESTS PASSING
01THE CHALLENGE
Run discovery, domain verification, ICP scoring, buyer discovery and email checks without duplicate work or uncontrolled request amplification.
02WHAT I WORKED ON
Built a 21-table PostgreSQL knowledge base and idempotent job pipeline; reduced request amplification from 42.8x to 4.12x with caching, redirect reuse and fetch guards, backed by 1,557 passing tests.
03ARCHITECTURE
01DISCOVER
External sources / user input
02NORMALIZE
Validation / dedupe / caching
03ORCHESTRATE
APIs / model calls / business rules
04SERVE
API / search / UI