What the work involves
You'll spend most of your time in two modes. In the first, you author reference material: a storyline for a renewal QBR, an executive summary that survives a CRO's ten-second skim, a data exhibit that makes a pipeline story land without a verbal walkthrough. In the second, you evaluate what the model produced against the same brief — marking where the horizontal logic breaks between headlines, where a slide has three ideas competing for the action title, where a chart choice buries the point it was supposed to make. Tasks are briefed in writing and submitted asynchronously, usually with a rubric and a required rationale for each judgment.
The distinction that matters here is between decks that are decorated and decks that are argued. Labs already have plenty of people who can make a slide look tidy. What's scarce is someone who can say precisely why a message flow collapses at slide four and what the correct sequence would be — and who can write that explanation clearly enough that it becomes training signal.
What the platform screens for
Ethos runs an AI voice screen before any task access. Expect it to press on specifics: which company, what deal sizes, who your decks were actually shown to, and how you'd restructure a named example. Generic consulting vocabulary — "MECE," "so-what," "pyramid principle" — recited without a worked example tends to read as thin. The screen is also checking whether you can critique in writing at a level of granularity a model can learn from, rather than saying a deck "feels off."
Logistics
- Fully remote and asynchronous; no fixed hours, no standups
- 5–20 hours per week, with room for more if you want it
- $100/hour is the observed rate for this engagement, not a guarantee — bands vary by task type and calibration performance
- Expect an unpaid or lightly paid calibration round before steady task flow begins