Find work

We are testing a chatbot that helps tenants understand their rental agreements. It must explain, never advise, and it must say when someone needs a lawyer. You get a six-point rubric and 20 answers we already graded, so you can calibrate before you start. Score 300 answers in our spreadsheet: accuracy, tone, and whether it crossed into legal advice. Add a one-line reason for every score below 4. No legal background needed, but you must read carefully and be consistent. We spot-check 30 of your scores against ours.

We are launching a pre-seasoned cast-iron pan and want six short vertical ads for Instagram and TikTok. We have product photos and a few clips of the pan in use; the rest can be made with AI video tools. Each ad needs a different opening line, captions burned in, and our logo at the end. Food and hands must look real; no warped fingers or melting steak. We pick two scripts out of six before you make anything.

We sell kitchen tools on Amazon US. Our titles are long and miss the words buyers actually type. Rewrite 60 listings: title, five bullets and backend keywords for each, within Amazon's limits. We would like to see five first so we can check the voice, then the rest.

Our clinics get the same questions every day: opening hours, insurance, what to bring, how to reschedule. All of it is in our 30-page help centre, which nobody reads. Build a chatbot for our website that answers only from those pages, links the page it used, and hands over to a person (our contact form) when it does not know. It has to work on phones. We use a standard website builder and can paste in a script. We will test it with our own list of 50 questions before we accept it.

Our heating and cooling business misses most calls after 6pm, and those are often emergencies. We want an AI receptionist that answers, works out if it is urgent, books a visit into our calendar or texts the on-call technician, and sends us a summary of every call. It must sound natural, spell names and addresses back, and never promise a price. You choose the platform; tell us the monthly running cost.

We have 40 photos of pans and utensils taken on a kitchen counter. We need them on pure white, square, with a soft natural shadow, ready for our store. Colours must stay true; please do not make the steel look blue.

When someone fills in our quote form, we want the contact in HubSpot, a two-line AI summary of what they need, and a Slack message to the right team (dental or home services) within a minute. You can use Zapier or Make; we have both. Please document the steps so our office manager can fix it if it breaks.

We are training a model to catch bad product photos before they go on our store. You will label 2,000 images in a simple web tool: pick the product category (12 options) and tick any defects (blurry, cut off, wrong colour, watermark, cluttered background). Most images take a few seconds. Full instructions and 40 worked examples are provided. Accuracy matters more than speed: we review a sample and ask for fixes if more than 3 in 100 are wrong.

We need 50 original word problems that today's models get wrong at least some of the time. Secondary-school to first-year-university level: rates, probability, geometry, a little number theory. Each problem needs one correct answer, a full worked solution, and a note on the trap (the step a model tends to skip). Nothing copied from textbooks or the web; we check. We will run each problem through two models before accepting it, and pay for every problem at least one of them fails.

We ran a webinar on caring for cast-iron pans. Find the 10 best moments (30–60 seconds each), reframe them to vertical, add captions and a title card. AI clipping tools are fine as long as you pick the moments yourself.

Our store chatbot tells customers we ship in 1–2 days. We ship in 3–5. It also invents a free-shipping threshold we do not have. Someone set it up last year and is gone. Find where it gets these answers, fix them, and make sure it says "I'll check with the team" rather than guessing when it does not know. Then show us 10 questions you tested it with.

We recorded 15 short videos showing how to use our product. Turn each into a help article: a clear title, numbered steps, a screenshot per step, and a "what if it doesn't work" section. Plain English for non-technical readers. We will review the first two before you do the rest.

Our front desk answers 80 emails a day. We want every new email to get an AI draft reply saved in Gmail, using our help pages and tone, that a person reads and sends. Nothing is ever sent automatically. Drafts must say when the AI is unsure. We will judge it on a week of real emails.

We are comparing two translation models on customer-support messages. For 400 English messages you will see two Vietnamese translations side by side. Pick the better one, score each from 1 to 5, and fix the worse one when it is below 3. You must be fluent in both languages. Southern and Northern usage are both fine; just tell us which you write.

We want our support team to search past tickets by meaning, not keywords: "customer can't log in after password reset" should find every similar ticket even when the words differ. Our stack is Postgres and TypeScript. Add pgvector, embed the 5,000 tickets, keep new ones embedded, and build a small search page with filters for date and product. Show us how good it is on 30 test searches we provide.

Half of our patients message us on WhatsApp, many in Spanish. We want a bot that books, moves and cancels appointments in our booking system, in either language, and passes anything medical to the front desk. Our booking system has an API and we have a WhatsApp Business account. You will build the flow, connect both, and train our front desk for 30 minutes on how to take over a conversation.

We are adding 12 scented candles to the range. We have clean product shots and want each one placed in a warm kitchen or dining scene that matches our existing photos (we will send 10 references). The candle and its label must stay exactly as photographed: no changed text or shapes.

Our app checks AI answers for mistakes. We need a 90-second demo for our homepage: problem, three features, one result. You record the screens (we give you a test account), write the script with us, and use an AI voice we approve. Clean and calm, not hype. Captions on.

We have 500 supplier invoices as PDFs, in about 12 different layouts. We need supplier, date, invoice number, total and tax in a sheet, and a way to run the same thing every month. Accuracy matters: flag any invoice you are not sure about rather than guessing.

Our refund bot is supposed to follow a strict policy: no refunds after 30 days, never promise money it cannot give, never reveal another customer's order. We want you to break it. Find 40 distinct prompts or conversations that make it break a rule. For each, write what you tried, what it said, and which rule it broke. Variety counts: 40 versions of the same trick count as one. You get a test account and the written policy. Please do not test anything outside the bot.

We already have a voice agent answering our booking line. Callers complain that it talks over them and gets names wrong. We have 40 recordings showing both problems. Listen, adjust its settings and prompt, and show us it is better on the same 40 situations. You will get access to the platform and the recordings.

Customers keep emailing to ask where their order is. Build a page on our Next.js site where they enter their email and order number and see the status and tracking link from our shop's API. It must not reveal anything without both fields matching, and must look right on phones.

We send a weekly email to 4,000 small-business owners about AI tools they can actually use. Write four issues: one tool tried properly each week, what it costs, what it is good and bad at, and one short tip. You must try every tool yourself; no rewritten press releases.

Before we change our support bot we want a fixed test set. Write 120 questions a real customer might ask (including rude, vague and misspelt ones), each with the answer the bot should give, taken from our help pages. We will share the help pages and 10 examples of the style we want.

Our pricing module (discounts, tiers, tax, currency rounding) has no tests and we are about to change it. Write around 40 unit tests that pin down how it behaves today, including the odd cases. If you find a bug, write a failing test and tell us; do not fix the code.

We have 10 hours of recorded calls with wholesale buyers. We need clean transcripts (an AI first draft is fine, but you must correct it) and tags for every objection, price question and competitor mention. US English, some background noise. We will send a tag list with examples.

We have 14 Zaps that someone built over two years. Five are failing, and nobody knows what the others do. Fix the broken ones, turn off any that are duplicates (ask us first), and write one plain paragraph per Zap.

We want a steady run of short ads that show our pans in use. Twelve 10–15 second ads over the month: quick recipes, a weekly deal, and one per week from our warehouse team. We will send phone footage every Monday. You edit, add AI shots where they help, and send post-ready files back by Thursday. If the month goes well we will make it ongoing.

We are releasing a small habit tracker. We need six App Store screenshots: phone frame, one short headline each, our colours. We will send screen captures and the copy; you can suggest better headlines.