Everyday AI Product Tester Needed
Budget / SalaryHourly project
TypeFreelance project
LocationRemote
Posted2 hours ago
Product Tester, Everyday AI
Remote | Part time | Independent contractor | Any time zone
Role: Product Tester, Everyday AI
Engagement: Independent contractor, part time
Location: Fully remote, any time zone
Commitment: Typically two to four sessions a week, scheduled by you
Reports to: A cohort lead
Pay: Hourly, by level. Stated before you commit.
AfterQuery is building a small family of AI assistants for the ordinary parts of a person’s week. We are looking for people to use them for real and tell us precisely what is wrong with them.
ABOUT THE WORK
Our assistants work on the recurring, unglamorous things: keeping a schedule and the commitments that go with it, planning meals and getting the shopping right, everyday money, errands and journeys, and the administrative post that has to be dealt with. They are narrow on purpose. None of them is a general assistant, and each one is meant to know where its own edges are.
We run a lot of automated evaluation, and it is good at what it does. It catches factual errors, broken output, crashes and regressions. What it cannot catch is an answer that is correct and still wrong: the right suggestion at a moment the person cannot act on it, a decision the assistant made that was the user’s to make, a product that is individually reasonable and collectively exhausting to live with. Those failures need a person who was actually there.
That is what this role is. You use the products the way their owners will, notice where they behave in a way a thoughtful person would not, and turn that into a report precise enough for an engineer to reproduce on a Tuesday morning without you in the room. Both halves are the job.
WHAT YOU WOULD BE DOING
- Running assigned test scenarios end to end and writing up what happened.
- Using an assigned product in your own life for a week at a time, and keeping a short daily log while you do.
- Probing deliberately for the failures that only show up in a life already in progress, rather than on a clean test account.
- Filing defects with reproduction steps, what you expected, why you expected it, and an honest statement of how sure you are.
- Checking that fixes actually hold, including saying when they do not.
- Rating recorded interactions against a fixed rubric, if you take that on. It is scheduled and paid separately.
- Writing a short note at the end of each four week cycle saying what the cycle actually established.
WHAT WE ARE LOOKING FOR
Observation kept separate from conclusion: You can write down what happened in a way that someone who disagrees with your conclusion would still accept as accurate, and then give the conclusion separately. This is the single strongest predictor of someone who does well here, and it is what our selection exercise is built to measure.
Clear ordinary writing: Reports are read by people who were not there. Plain English is enough. Polish is not assessed and long reports are not better ones.
A life the products touch: You keep a calendar, or you cook, or you commute, or you deal with official post. We match assignments to this, because the field weeks depend on it being genuine.
Reliability at low volume: Two sessions a week that actually happen is a better engagement than five that slip.
Willingness to be unimpressed: Including with us. We need people who will say a product is not good enough, and equally people who will say an area is fine so that attention goes where it is needed.
WHAT YOU DO NOT NEED
No software testing background. No QA certification or experience with formal test methodology. No coding, no reading code, and no understanding of how the models work. No technical degree. On some assignments it is better if you have never used a product like these before.
The bar is judgment, not credentials. Our strongest testers have come from teaching, nursing, translation, hospitality management, accountancy and freelance writing. What they had in common was the habit of noticing when something is slightly off, and the discipline to write it down precisely.
TERMS
Engagement: Independent contractor, part time, renewed by four week cycle. No minimum total term.
Location: Fully remote, any time zone. Some assignments include work away from a desk, in a shop or on a journey.
Hours: Typically two to four sessions a week of ninety minutes to two hours, scheduled by you. No standing meetings and no required hours of the day.
Pay: Hourly against recorded time, at one of three levels. The rate that applies to you is stated before you are asked to commit to anything. Onboarding, reading, and the selection exercise are all paid.
Equipment: A current computer and phone. Nothing to buy. Where an assignment needs hardware you do not have, we lend it.
Responsiveness: One working day for ordinary messages. There is one short list of safety matters that need a response within the hour, and nothing else in the role is urgent.
HOW TO APPLY
1. A short form. Where you are, roughly what your week looks like, and which of our domains your actual life touches. A CV is accepted and is not read closely. There is no cover letter.
2. A paid exercise, under an hour, in your own time. You are given a recorded interaction with an assistant and asked to write it up. There is no single correct answer and there are several ways to do well.
3. A conversation of about forty minutes, mostly about your exercise. No panel, no second round, no presentation.
The whole process takes about two weeks. We do not ask for unpaid work at any stage. If the exercise takes you longer than an hour, stop and send what you have; that is a real instruction and not a test.
ONE THING WORTH KNOWING FIRST
The work is quiet, detailed and repetitive in places, and its pleasures are small and specific: the moment you can name what was wrong with something that merely felt wrong, and the moment a product changes because you wrote a paragraph carefully. People who enjoy that tend to stay for many cycles. If you are looking for a route into a different role in AI, this is honestly not one, and we would rather say so now than have you find out in three months.
AfterQuery hires without regard to background, and we assess the exercise before we look at anything else in an application. If you need an adjustment at any point in the process, ask, and it will not form part of the assessment.
Remote | Part time | Independent contractor | Any time zone
Role: Product Tester, Everyday AI
Engagement: Independent contractor, part time
Location: Fully remote, any time zone
Commitment: Typically two to four sessions a week, scheduled by you
Reports to: A cohort lead
Pay: Hourly, by level. Stated before you commit.
AfterQuery is building a small family of AI assistants for the ordinary parts of a person’s week. We are looking for people to use them for real and tell us precisely what is wrong with them.
ABOUT THE WORK
Our assistants work on the recurring, unglamorous things: keeping a schedule and the commitments that go with it, planning meals and getting the shopping right, everyday money, errands and journeys, and the administrative post that has to be dealt with. They are narrow on purpose. None of them is a general assistant, and each one is meant to know where its own edges are.
We run a lot of automated evaluation, and it is good at what it does. It catches factual errors, broken output, crashes and regressions. What it cannot catch is an answer that is correct and still wrong: the right suggestion at a moment the person cannot act on it, a decision the assistant made that was the user’s to make, a product that is individually reasonable and collectively exhausting to live with. Those failures need a person who was actually there.
That is what this role is. You use the products the way their owners will, notice where they behave in a way a thoughtful person would not, and turn that into a report precise enough for an engineer to reproduce on a Tuesday morning without you in the room. Both halves are the job.
WHAT YOU WOULD BE DOING
- Running assigned test scenarios end to end and writing up what happened.
- Using an assigned product in your own life for a week at a time, and keeping a short daily log while you do.
- Probing deliberately for the failures that only show up in a life already in progress, rather than on a clean test account.
- Filing defects with reproduction steps, what you expected, why you expected it, and an honest statement of how sure you are.
- Checking that fixes actually hold, including saying when they do not.
- Rating recorded interactions against a fixed rubric, if you take that on. It is scheduled and paid separately.
- Writing a short note at the end of each four week cycle saying what the cycle actually established.
WHAT WE ARE LOOKING FOR
Observation kept separate from conclusion: You can write down what happened in a way that someone who disagrees with your conclusion would still accept as accurate, and then give the conclusion separately. This is the single strongest predictor of someone who does well here, and it is what our selection exercise is built to measure.
Clear ordinary writing: Reports are read by people who were not there. Plain English is enough. Polish is not assessed and long reports are not better ones.
A life the products touch: You keep a calendar, or you cook, or you commute, or you deal with official post. We match assignments to this, because the field weeks depend on it being genuine.
Reliability at low volume: Two sessions a week that actually happen is a better engagement than five that slip.
Willingness to be unimpressed: Including with us. We need people who will say a product is not good enough, and equally people who will say an area is fine so that attention goes where it is needed.
WHAT YOU DO NOT NEED
No software testing background. No QA certification or experience with formal test methodology. No coding, no reading code, and no understanding of how the models work. No technical degree. On some assignments it is better if you have never used a product like these before.
The bar is judgment, not credentials. Our strongest testers have come from teaching, nursing, translation, hospitality management, accountancy and freelance writing. What they had in common was the habit of noticing when something is slightly off, and the discipline to write it down precisely.
TERMS
Engagement: Independent contractor, part time, renewed by four week cycle. No minimum total term.
Location: Fully remote, any time zone. Some assignments include work away from a desk, in a shop or on a journey.
Hours: Typically two to four sessions a week of ninety minutes to two hours, scheduled by you. No standing meetings and no required hours of the day.
Pay: Hourly against recorded time, at one of three levels. The rate that applies to you is stated before you are asked to commit to anything. Onboarding, reading, and the selection exercise are all paid.
Equipment: A current computer and phone. Nothing to buy. Where an assignment needs hardware you do not have, we lend it.
Responsiveness: One working day for ordinary messages. There is one short list of safety matters that need a response within the hour, and nothing else in the role is urgent.
HOW TO APPLY
1. A short form. Where you are, roughly what your week looks like, and which of our domains your actual life touches. A CV is accepted and is not read closely. There is no cover letter.
2. A paid exercise, under an hour, in your own time. You are given a recorded interaction with an assistant and asked to write it up. There is no single correct answer and there are several ways to do well.
3. A conversation of about forty minutes, mostly about your exercise. No panel, no second round, no presentation.
The whole process takes about two weeks. We do not ask for unpaid work at any stage. If the exercise takes you longer than an hour, stop and send what you have; that is a real instruction and not a test.
ONE THING WORTH KNOWING FIRST
The work is quiet, detailed and repetitive in places, and its pleasures are small and specific: the moment you can name what was wrong with something that merely felt wrong, and the moment a product changes because you wrote a paragraph carefully. People who enjoy that tend to stay for many cycles. If you are looking for a route into a different role in AI, this is honestly not one, and we would rather say so now than have you find out in three months.
AfterQuery hires without regard to background, and we assess the exercise before we look at anything else in an application. If you need an adjustment at any point in the process, ask, and it will not form part of the assessment.
Apply on Freelancer →
Project sourced from Freelancer.com. Applications happen directly on the original platform — we never collect your data.