What the work actually involves

You take a contract scenario — often built around an APA, SPA, NDA, or commercial agreement in the tech space — and work it the way you would for a client: mark up the draft, propose alternative language, and articulate why a given indemnity cap, materiality qualifier, or closing condition is or isn't acceptable from a named party's position. A second strand of the work is grading: you read what a model produced against the same scenario and judge whether its markup reflects real transactional judgment or merely plausible-sounding contract prose. A third strand is framework design — writing the grading rubric itself, so that another attorney scoring the same output would reach roughly the same conclusion.

The unit of work is the task, not the hour. Average handling time is roughly 3.5 hours, and you are paid per task that meets spec and is signed off. That means rework on a rejected task is uncompensated time, so reading the annotation guidelines closely at the start of a project matters more than raw speed. The $90–130/hr band is an observed effective rate across contributors, not a guaranteed wage.

What the platform screens for

  • Genuine deal-side experience. Expect probes on specific M&A mechanics: purchase price adjustments, sandbagging provisions, escrow and holdback structures, reps survival periods. Vague familiarity surfaces quickly under follow-up.
  • Reasoning you can externalise. A redline is worth little to a training pipeline without the rationale. Screeners look for whether you can state the risk, the party bearing it, and the negotiating posture behind your edit.
  • Consistency under a rubric. Can you grade against stated criteria rather than against your personal drafting preferences? This is where many strong practitioners lose points.
  • Bar standing and JD provenance, which are verified.

Logistics

Fully remote, contractor, part-time, and largely asynchronous — tasks are claimed from a queue rather than scheduled. Most contributors work in blocks, since a single task rarely splits cleanly across sittings. Occasional calibration calls with research or product teams occur during US business hours. There is no minimum commitment, but throughput tends to determine whether you stay on a project as it scales.