What the work involves
You are the caller on the other end of a customer-service AI. Mercor's client needs Mexican Spanish speakers to run through scripted and semi-scripted scenarios by phone — asking about an order, changing an appointment, disputing a charge, interrupting mid-sentence, switching topics — and then report what the agent did with it. Each session is a handful of calls plus a short written or form-based debrief: did the agent understand the request, did it handle regionalisms and filler speech, did it hallucinate a policy, did it recover after being cut off, did latency make the exchange feel unnatural.
There is no domain expertise required and that is deliberate. The client wants ordinary speakers behaving like ordinary customers, not linguists performing test cases. What matters is that you speak Mexican Spanish natively or near-natively, that you can be understood on a phone line, and that your feedback distinguishes "the agent misheard me" from "the agent heard me and answered wrong."
What the screen looks for
- Genuine Mexican Spanish, not general Latin American Spanish — usted/tú handling, local vocabulary, natural speed and elision
- Call quality you can actually produce: a quiet room, a reliable phone or headset, no cafés or shared spaces mid-call
- Availability now, in defined daily blocks, for a two-week window — this project ends, and flakiness in week one ends it early
- Feedback that is specific: timestamps, quotes, what you said and what came back
Logistics
Fully remote and phone-based. Sessions are typically arranged in overlap with the client's business hours, so expect some scheduling constraint rather than fully async work. Commitment is 1–3 hours per day for 1–2 weeks with an immediate start; pay is observed at $10/hr and is not guaranteed by Sidequest. Renewal beyond the initial window depends on the client, not on your performance alone.