Gmail & Google Calendar AI Assistant Evaluator

What you'll do

The tasks are the ones people already delegate to an assistant: answer this thread, move that meeting, book a table for four, find a time that works for three people and send the invite. You give an AI assistant clear instructions for tasks like these, then check its work the way you would check a new assistant's, looking at whether the email went to the right person, whether the invite landed on the right day and whether the booking actually exists.

Every run happens under standardized, repeatable conditions, so the same scenario can be compared across assistants and model versions. After each one you rank and annotate the responses against the project's grading guidelines and log the results in micro1's tooling. The scenarios deliberately cover email threads, calendar invites, online bookings and multi-person coordination, including the awkward edge cases (a double booking, a reply-all that should not have been, a time zone mix-up).

There is also a team element: you work through ambiguous cases with project trainers and other contributors and take part in quality reviews, so expect your grading to be checked and calibrated.

Who fits

This is one of the few listings where the qualification is your ordinary life rather than a profession.

• You use Gmail and Google Calendar daily for your own affairs: sending mail, accepting invites, scheduling and rescheduling
• You regularly coordinate with other people over email or text (meetups, follow-ups, confirming shared plans or payments)
• You book things online without help: restaurants, appointments, travel, deliveries
• You already use AI tools for practical tasks and are comfortable instructing one and checking its output
• Careful attention to detail, clear written reporting and the discipline to work alone remotely

The hard requirements are unusual: you must be based in the US, own an iPhone with iMessage, and have a Facebook or Instagram account. The ad does not explain why, but the likely reading is that some test scenarios run through those apps, which would put your personal accounts in the loop. Prior AI evaluation, data labeling or user testing, prompt writing and grading AI output for accuracy, helpfulness and safety are preferred rather than required.

What it pays

$15–30/hour, micro1's published band. That is entry-level money for AI evaluation work, which fits a role whose main credential is being a competent everyday user. The ad does not say what decides placement in the band, how many hours are available or how long the project runs.

Worth knowing

Good:

• No degree, profession or years of experience required
• 50 openings, with a "High demand" badge
• The skill being tested is one most organised people already have
• Prior evaluation or user-testing experience is a plus, so experienced raters have an edge

Less good:

• US residents only, and only with an iPhone plus a Facebook or Instagram account
• Tasks may touch your own email, calendar or social accounts; ask during screening whether test accounts are provided before you connect anything personal
• Bookings made during testing could carry real costs or cancellation fees; the ad says nothing about who covers them
• Low ceiling for the category
• No hours, duration or start date published
• Contractor engagement: no benefits, your own tax to manage

About this listing

Posted by micro1 on its own jobs portal on 25 September 2026 and read the next day. The pay band, the 50 openings, the US-only location and the iPhone, iMessage and Facebook or Instagram requirements are micro1's own text for this listing. The customer behind the project is not named. See micro1.