Dave Jones founded MainThread in 2026 and built Genesis as its first production platform. He is the operator on every candidate account — the person designing each bespoke harvest, reviewing every intelligence briefing before publication, running the direct-messaging thread, and shaping the judgment AI alone cannot.
The cohort size is intentional. Genesis onboards roughly eight candidates a month because calibrating a harvest around someone properly — title vocabulary, stage range, company types, off-limits cells, comp floors — takes real time. Batching keeps the work deep instead of thin.
Dave is also his own Pattern Zero. His active harvest runs alongside every candidate's. The patterns named on his map are frequently the ones that first surface a phenomenon the rest of the cohort later confirms. The architect stays a candidate.
Genesis runs on four architectural commitments. These aren't features; they're the shape of the work. Every shipped capability is one of these four principles made visible.
The harvest starts candidate-specific.
Every candidate's title vocabulary gets mapped before the first harvest runs — typically 20–40 variants the market actually posts under. The pipeline seeds against those variants across 6,000+ career pages. When the candidate explores a new niche, the harvest extends to catch it. Precision starts at configuration, not at filter time.
The map compounds across every briefing.
Named Market Patterns — Pharma Velocity, the Comp Cliff, the AI-Native Ops Stampede — are first-class objects that accumulate evidence across briefings. Emerging → active → sunset. The candidate never starts cold; each briefing extends a living map of their market. No other AI job search tool has named longitudinal memory as a category axis.
AI scale, operator judgment.
Multiple AI research lenses process hundreds of roles per briefing in parallel. The operator designs the harvest, reviews every briefing before publication, and shapes the judgment AI alone cannot. Not pure automation. Not pure service. Two intelligences coordinating on the same candidate.
Every signal sharpens the next harvest.
Swipes, conversations, interviews, beacon responses — each one bends the arc of what comes next. The harvest widens when the candidate explores; the rank tightens when the candidate decides. The feed gets sharper every day because the intelligence layer is listening in both directions.
Each briefing contributes to a living map. Each candidate's named patterns feed the methodology the next candidate inherits. Here is where Genesis stands today.
Release notes at /changelog. Agent-discovery context at /llms-full.txt.
The harvest gets shaped around one trajectory at a time.