An AI assistant for long documents, analysis and code, with Claude Code and Cowork on paid plans.
Freemium
AI trainer and data annotator
New roleRates and corrects model answers, writes reference solutions and labels the data that models learn from. Companies that train models increasingly look for people with subject-matter expertise, not just labelers.
Where this job sits on the map
In the same group: 14 of 100 professions.
What changes
Instead of simple labels on images, the work is more often judging a complex model answer and explaining what's wrong with it.
Subject expertise is valued: medicine, law, math, programming, rare languages.
The work is usually project-based and remote, through intermediary platforms, and the volume of tasks can swing sharply.
AI drafts, the person decides
How the work splits here: AI prepares a draft, the person checks it and makes the call.
What AI does
- Does a first pass of labeling that a person then checks
- Generates answer variants to compare
- Finds duplicates and obvious errors in the data
- Checks finished work against the instructions and flags the disputed spots
You check itRead it, check the facts, fix it. Without this check the draft goes nowhere.
What stays with the person
- Expert judgment: which answer is correct and why
- Staying careful across hundreds of similar tasks under a long set of instructions
- Reference solutions that a model can't write itself
- Turning down tasks that break the law or someone else's privacy
What to learn first
6 lessons from the course, in order. Start with the first one.
- Your first 30 minutes with AI: where to startStarting from zeroYour first 30 minutes with AI: sign up, ask three questions, try seven everyday tasks, learn what never to share and which plan to pick.Start here
- How an LLM works inside, explained without mathStarting from zeroHow a language model works inside: tokens, context, temperature and hallucinations, explained without math.
- How to write a good promptUserHow to give AI a task: the five parts of a good prompt, Plan Mode, and how to refine an answer instead of starting over.
- AI ethics and safety: hallucinations, attacks, biasUserHallucinations, prompt injection, bias, privacy and copyright: how not to trust AI blindly.
- Fine-tuning: when prompts aren't enoughBuilderWhen a prompt is enough, when you need RAG and when you need fine-tuning: a decision tree, LoRA and where to train.
- Evals: skills that improve themselvesEngineerEvals for skills: tests, pass rate, checking that the skill triggers and a data-driven improvement loop.
Which tools to use
OpenAI's mainstream AI assistant: chat, voice, images, the Work agent and Codex.
FreemiumGoogle's assistant, built into Gmail, Docs and Sheets, with Deep Research, Gemini Live and Canvas.
FreemiumAI search with sources: answers with links, Deep Research and the Comet browser.
Freemium
Ready-to-use materials
- How to check an AI answerChecklist
Assess the risk, spot the signs of a made-up answer and run a 5-minute check before you act on what AI says.
Free Go through an AI answer claim by claim: what's right, what's doubtful, where it's wrong and what to check at the source.
Free- What never to send to AIChecklist
A 30-second check before you hit send: what to remove, how to anonymize and what to do if it's already gone out.
Free
How to earn with AI
We don't promise income: results depend on your niche, your market and your work.
- Job
If you're a specialist in your field, take projects rating model answers in your specialty, and learn where AI goes wrong in your profession along the way. A platform that asks you to pay for access to tasks should raise a red flag.
- Service
Preparing test questions and reference answers for companies that are launching an assistant in your industry.
- Product
A set of test questions with reference answers for one profession, which can be used to check any assistant.
Similar professions
Checked: October 2026