What the work involves
You'll spend most of your time on three loops. First, authoring: you write reference decks and slide sequences from a prompt — a partner QBR, a co-sell business case, an exec summary for a revenue-share proposal — the way you'd build them for a real counterparty. Second, evaluating: the model produces a deck and you grade it against a rubric covering storyline, horizontal and vertical logic, exhibit choice, and whether the action titles actually carry an argument. Third, correction: you rewrite the weak slides and explain, in writing, why the original failed. That written reasoning is usually worth more to the lab than the fixed slide itself.
Typical task domains mirror partnerships work: storyline development, executive summary writing, slide structuring, data-driven exhibit design, and client-ready production quality. Expect to be asked why a waterfall beat a stacked bar, why an exec summary buried the ask, or why a deck that reads fine top-to-bottom falls apart when you test whether the sub-points actually sum to the title.
What the platform screens for
Ethos runs an AI voice screen. It is looking for a real partnerships track record at a high-growth tech or enterprise B2B company — named companies, named deal types, what you owned — and for the ability to critique a deck in structural language rather than aesthetic language. Candidates who say a slide is "cluttered" score worse than candidates who say the slide has four competing messages and no action title. Expect follow-ups that push a level deeper than your first answer; the screen is testing whether your craft vocabulary is load-bearing or borrowed.
Logistics
- Fully remote, asynchronous, work on your own schedule
- 5–20 hours per week, more if you want it; no fixed shifts
- $100/hour observed for this listing — rates on eval platforms vary by task type and can change between cohorts
- Expert-level PowerPoint is assumed; you will be producing files, not just commenting on screenshots