About AnnoTruelabel: founder · stefan

AnnoTrue is a data labeling company run by one person: me, Stefan. Here is why it exists and why that matters for your data.

How I Worklabel: principles

1

Fair Pay, Published

At least $10 an hour, or the legal minimum where the annotator lives if higher. Rushed, underpaid people produce rushed data.

2

Calibration First

Guidelines, a gold set and a calibration round, so disagreements surface as soon as possible, not in your eval results.

3

Quality You Can Check

Hidden gold items, overlap and reviewer sampling, with an accuracy report in every delivery.

4

Clear Pricing

You see what you pay and what annotators earn. No hidden fees, no minimums.

5

A Free Pilot

Judge the data, not the pitch.

Me and the Big Platforms: An Honest Comparisonlabel: fair

What I Do Better

  • One accountable person. You talk to the person who is responsible for everything and who reviews your data.
  • Guidelines that don’t drift. Every change is dated and announced to every annotator at once.
  • You come first. For a young company, every client shapes its name. Your project gets the care a large firm keeps for its biggest accounts.
  • Numbers you can verify. A price you can simulate, published pay and an accuracy report with every delivery.
  • No lock-in. A free pilot, no minimums, month to month.

What the Big Platforms Do Better

  • Scale. Thousands of annotators on a project overnight. I can’t, yet.
  • Breadth. Many languages and specialist fields. I mainly work in English.
  • Track record. Years of large AI-lab projects. I am building mine.

If you need thousands of annotators next week, they are your best choice. If you need a dataset you can trust, with one person accountable for every label, start with a pilot here.

A modellabel: belief is only as good as the judgment inside its training data.

That judgment should be measured, not assumed.