Guide · first-hand

What happens after you apply for AI training work?

By Alex, the person behind AIPayList. Published 31 August 2026.

If you want to become an AI trainer, this guide covers the practical path from application to paid work. It explains what happens after you apply, how platforms screen applicants, and where to browse live listings on our jobs board and platform profiles.

I work in operations and strategy. I have applied to micro1, Mercor and Ethos, worked on projects, and received payment from all three. Here is what I wish I had known before starting.

My main advice: apply to several roles that connect to your real experience. Do not choose one and wait. An imperfect match can still be worth an application, provided you meet the role's essential requirements.

This guide describes my experience, not a promise about your next project. Published platform rules have source links.

Prepare your CV and your opening answer

Every interview I took used an AI interviewer and opened with a request to describe my experience. Prepare that answer: what you do, which experience matters for this role, and a result or example that makes it concrete. Same career, different emphasis for each application.

I tailored my CV with Claude. I supplied my real CV and the job description, then asked it to interview me about relevant experience I had omitted. This helped me recall examples and explain details without inventing qualifications. Here is the prompt I use, free to copy.

I preferred the AI interview to speaking with a recruiter. I could choose when to start and had room to explain my examples and reasoning without worrying about giving the interviewer a turn. Stay focused on the question, but give your experience enough space.

Expect exercises, screen sharing and occasional technical problems

Some interviews include a timed exercise, and it is not a sample of the work itself. In my experience it was a business scenario that asked for my reasoning, typed into text boxes inside the time allowed.

In one, I was given a business with a profit and loss statement and some context, and asked to assess it as a management consultant would: apply a MECE structure, identify the root causes, and propose how to improve the situation. In another, I had to explain how I would build an executive-level presentation on a given topic, proposing the storyline, the outline of the slides, the data I would use and the charts I would pick.

Those are illustrations of the kind of thinking asked for, not the prompts themselves, and what you get will depend on the role. I encountered exercises of around 25 minutes on both micro1 and Mercor.

They always checked that my microphone and camera were working, then asked me to share my screen, which is there to stop people cheating. Have all three ready before you start. Then follow the assessment's own rules on assistance. Mercor restricts AI-generated substantive interview answers, and when a Mercor interview requires screen sharing you must share the full screen rather than a single window. Mercor interview guidance

Technical problems were rare for me. When an interview failed, I closed it, returned to the role and restarted the unfinished step without support. That is not a promise of unlimited attempts, so check the platform's recovery rules. Mercor's retake limits and support options

Applying to more relevant roles has become easier

When I started, I kept hearing the same questions on later applications. Interview reuse has since reduced that repetition. Another application can now take much less effort than the first.

On Mercor, many later applications needed no new interview from me. With the required assessments complete and no other outstanding steps, applying could be close to one click. Mercor documents reuse of compatible assessments, not blanket qualification for every role. Mercor application guide

On micro1, reuse was often partial. I still answered questions about one to three additional skills. Some follow-up interviews took around 8 to 12 minutes instead of the 20-plus minutes I remembered from earlier applications.

I once applied to work that overlapped with my experience but felt more specialized than my generalist background. I answered from experience, explained my reasoning, and was accepted. Relevant experience does not always come with the exact job title in the listing.

Applying to many roles did not stop me getting work. Fit and availability still matter, and more applications alone do not guarantee acceptance. Mercor's interview guidance also encourages further applications while waiting for a response.

I would act when a suitable role appears and reserve enough uninterrupted time to complete the interview. Listings can close, but speed alone does not explain selection. I have had answers within a day or two and waits longer than a month.

Browse current AI work or set up matching job alerts to spend less time checking job boards.

Separate the advertised rate, your request and the offer

A listing's maximum is not a promised payment, and a requested rate is not an offer. Read the eventual terms before deciding whether the work suits you.

On micro1, I entered a requested rate for the job. One listing showed about $180/hour. I requested about $160/hour and received the rate I entered. Those amounts are approximate recollections. This describes my application, not proof that a lower request caused acceptance.

On Mercor, the application I checked did not ask for a rate. Work Preferences had separate full-time and part-time entries, including a minimum expected amount in dollars per hour for part-time work. My offer matched my minimum. That does not establish that Mercor always caps higher offers at your preference. Mercor's documents distinguish opportunity preferences from contract terms. Pay preferences, Mercor contracts

For Ethos, confirm the offered rate and pay unit for the particular role. The micro1 and Mercor examples above do not establish a shared rate-setting process for all three platforms.

How I would choose a requested rate

Where I can request a rate, I start with an amount I would be happy to accept. For a strong specialist match, or when I can wait, I would consider the upper end of the advertised range, sometimes around 5 to 10 percent below the maximum. This is my starting point, not a proven advantage over other applicants.

For broader work where I wanted a first opportunity, I might consider a comfortable middle-of-range request. If applications produced nothing, I would review the fit, CV, assessments and rate together. The lowest rate is not my target.

Use the pay index to compare advertised ranges and read how it is calculated. Listing ceilings and counts are not typical earnings, active tasks or acceptance odds.

What onboarding and monitoring were like for me

After acceptance came onboarding and setup. These timings are estimates from my projects, not platform averages or promised paid allowances.

I recall paid onboarding and training time on my projects, but not a universal allowance. Before starting, ask which setup, training, webinars and trial tasks count as paid time. Check when to start logging and whether paid training has a cap. Do not assume every step before production work is billable.

Understand what the tracker records

On my micro1 project, Hubstaff showed time and mouse or keyboard activity, not screenshots, app names or website history. That describes what I saw, not every possible setup. Hubstaff supports configurable monitoring features beyond time tracking. Hubstaff's tracking explanation

Mercor required Insightful. I observed app and website usage monitoring and felt less comfortable. I am unsure which screenshot settings applied to my project. Mercor's time-review documentation refers to screenshots, so check the configuration before assuming a tracker records time alone. Mercor time and contract guidance

Start well, check your hours, and keep expectations realistic

I would repeat three habits: finish onboarding without unnecessary delay, start the approved production tasks, and focus on quality. On Ethos, I received praise and an invitation to another project. That was encouraging, not a promise of automatic progression.

Check weekly commitments before accepting. I could work more than 20 hours on one micro1 project, and my Ethos project had no minimum in my experience. Mercor involved confirming my availability. These arrangements can differ, so confirm the minimum commitment, maximum hours, start date and flexibility during onboarding.

Ethos once flagged a task of mine for taking more time than expected, then removed the flag. I do not know the reason for its removal. My approach is to record time honestly, follow instructions, and ask what needs correcting if a flag appears.

One thing worth knowing about Ethos before you start: their training is paid. On my projects that meant roughly three hours of training at the project rate, which is not how most platforms handle it, and it is genuinely in their favour. Unpaid onboarding is common elsewhere.

On my projects across these three platforms, I was paid for hours logged, once every 14 days. I have not experienced late or incorrect payments. That was my arrangement. Mercor's current documentation also covers task-based work and describes weekly payout cycles, so check your own offer. Mercor contract terms

Ethos paid me through Wise, in dollars, so a Wise account that can receive dollars is the simple route. If you do not have one yet, my Wise invite gives you a free Wise card or a fee-free first transfer, whichever Wise offers in your country.

Before accepting, confirm the pay unit, schedule, paid activities and task-acceptance conditions. Logging hours against a task is not the same as agreeing a fixed fee for that task.

Projects can be short. One of mine lasted a little over a month. A high advertised rate does not tell you how many paid hours you will get or how long the work will last. Acceptance does not guarantee steady work.

How I would start again

1. Prepare a master CV with examples I can explain.
2. Set relevant job-alert filters and tailor the CV for suitable roles.
3. Prepare for interviews and check which assessments already count.
4. Review any offer and confirm I can meet the workload and start date.

I value what I learn. Reviewing AI output and evaluation criteria has shown me how projects use human expertise and how those needs change. That interests me beyond the pay.

Repeated applications without an offer are frustrating. I cannot promise the next one will succeed. Keep looking for roles connected to your experience, make that connection clear, and judge opportunities by their actual terms.

Get matching roles without checking every board

Would a future interview-preparation newsletter help? Tell me what you would like covered at contact@aipaylist.com. Job alerts are available now, and a separate tips newsletter has not launched.

Compare these platforms on pay:Mercor vs micro1AfterQuery vs MercorAlignerr vs MercorDataAnnotation vs micro1micro1 vs Outlier (Scale AI)

All guidesHow to get these jobsThis week's pay

Disclosure: some outbound links are referral links. See the referral disclosure. Found something different? Email the same address with a non-confidential description so I can check the guide.