What the work involves

You will spend most of your time doing two things: producing reference-quality decks against a prompt, and judging what the model produces against that same prompt. A task might be a quarterly earnings storyline for a mid-cap industrial, a two-slide executive summary for a board deck, an analyst-day section on segment margin drivers, or an exhibit that turns a messy cash flow table into something a portfolio manager reads in eight seconds. Alongside the artifact, you write the reasoning — why this storyline order and not another, why the horizontal logic across a five-slide section holds or breaks, where an action title asserts something the chart underneath does not actually support.

Evaluation work is where the judgment shows. Models produce decks that look plausible and fall apart structurally: action titles that describe rather than argue, sections whose vertical logic doesn't ladder to the governing message, exhibits that chart the wrong unit of analysis. You will be asked to rank outputs, mark the specific failure, and write a correction a non-expert reviewer can follow.

What the platform screens for

  • Manager-level ownership, not support. Screens probe whether you owned the narrative and the deck end to end — earnings call materials, analyst days, investor presentations — or assembled slides someone else structured.
  • Structured communication vocabulary used correctly. Expect follow-ups on storylining, horizontal and vertical logic, MECE grouping, and action titling. Loose usage is noticed.
  • Concrete disclosure instincts. Whether you can say what belongs in an IR deck, what needs legal review, and what a forward-looking statement can and cannot claim.
  • Slide craft specifics. PowerPoint fluency at the level of exhibit construction, not template filling.

Logistics

Fully remote and asynchronous — tasks are claimed from a queue and completed on your own schedule, typically in one- to three-hour blocks. Commitment is flexible at 5–20 hours per week, with more available. The interview process includes an AI voice screen. The $100/hour rate is as observed on the platform and is not guaranteed; rates vary by task type and reviewer tier.