Frequently Asked Questions

If you can't find the answer to your question,
feel free to contact us.

Platform

Do you provide a data labeling workforce?

Yes, our data labeling platform provides labeling tools and an elite data labeling workforce in 30+ languages. We believe that finding high-quality labelers at scale with the skills needed for your particular tasks is one of the most challenging aspects of data labeling. So unlike other data labeling companies, we don’t outsource our workforce – we fully manage it ourselves, which means higher quality and lower costs.

Do you provide labeling tools?

Yes, we provide rich labeling tools in addition to our high-quality, large-scale workforce. Our WYSIWYG editor allows anyone – even non-technical members of your team – to easily create data labeling templates to show to labelers. These fully customizable data labeling templates allow you to define instructions, present your data to labelers, and create as many multiple-choice, checkbox, and free response questions as you want. You can also create bounding boxes, named entity recognition tagging, relationship tagging, conversational chatbot dialogues, and more.

How many questions can I add to a template? Are there any limits to the complexity of a task?

As many as you want. There are no limits. You can combine image questions with text questions with free response questions with open-ended research. Some of our customers run “tasks” that can individually take hours, like writing long article summaries to train summarization algorithms, researching misinformation for content moderation, or coding Python programs to train neural synthesis programs.

Do you have an API?

Yes, our API allows you to programmatically create, monitor, and retrieve projects, and is interchangeable with our WYSIWYG frontends.

Can I self-serve?

Yes, our self-serve platform means that anybody can get started labeling in minutes, with no committed spend and no annual platform fees. Simply upload your data, create your questions in our WYSIWYG editor or API, add your credit card, and Surgers will start labeling.

Do you provide a managed service?

Yes, we also offer a fully managed service for companies and research labs with more complex needs. We’ll help you write labeling guidelines, build custom data labeling teams, recruit for any skills you need, design custom quality controls, help you manage volume and quality SLAs, and provide a dedicated project manager and 24/7 support.

Reach out to team@surgehq.ai for more details.

I have my own data labelers. Can I use your tools without your labeling workforce?

Yes, you can also bring your own labelers onto our data labeling platform, with no cost. Reach out to team@surgehq.ai for more details.

Can you help me write labeling instructions?

Yes, we’re always happy to help you create great labeling guidelines.

What kinds of quality controls do you provide?

We have a powerful suite of quality controls, including gold standards, outlier detection, AI-assisted labeling, custom data labeling teams, multiple labelers, dynamic matching, and more. Reach out to support@surgehq.ai for more details.

Do you perform data collection too?

Yes, we can also help you gather data from scratch, including audio collection, video collection, photography collection, and more. Reach out to support@surgehq.ai for more details.

Can I build my own data labeling team that stays with me over time?

Yes, you can build as many custom data labeling teams as you’d like, and use them or combine them in your projects.

Can I communicate with my data labelers?

Yes, our platform provides robust chat and messaging controls.

Can I give labelers an exam?

Yes, you can give labelers an exam they must pass before they can work on your projects. Surgers who perform particularly well can also be added to a custom data labeling team.

Do you support real-time data labeling?

Yes, many of our customers run projects where they require results to be returned in seconds or minutes. If you have low-latency SLAs, reach out to support@surgehq.ai to discuss more.

Data Labeling Workforce

What languages do you support?

We support 30+ languages, including: English, Spanish, French, German, Japanese, Portuguese, Arabic, Korean, Mandarin, Turkish, and Amharic.

If you need extra language support, reach out to support@surgehq.ai.

Tell me more about your data labelers

Unlike typical data labeling companies and BPOs that outsource their data labeling to low-skilled labelers in third world countries who have minimal command of English or other languages, we build our workforce out of highly skilled and educated native speakers.

For example, our Surgers come from all walks of life – whether they’re stay at home parents, teachers looking to earn extra money after school, Ivy League students paying their way through college, or software engineers interested in advancing AI.

What skills and backgrounds do your data labelers have?

Our personalization engine matches Surgers with the skills your data labeling projects need. This means that not only do we use native English speakers for your English labeling projects (as crazy as this sounds, this is rare in the data labeling world), native Spanish speakers for your Spanish labeling projects, and so on – but we also use Surgers with accounting and finance backgrounds for your financial categorization projects, actual Twitter users for your tweet labeling projects, gamers for projects labeling gaming messages, software engineers and CS grads for training coding AI assistants, and so on.

If you need special skills and backgrounds, just ask!

Pricing

How much does it cost?

Pricing works on a per-task basis. We don’t charge a platform fee.

If you self-serve your own projects, you choose your own price. In order to ensure a fair wage to labelers, we suggest trying to price your tasks at a $20-25/hour rate. This means, for example, that if your task takes a minute on average to complete, then it should be priced at $0.35-$0.40. If you need help with pricing, just reach out to support@surgehq.ai.

If you use our fully managed service, we’ll help price each task using the same methodology.

See our pricing page for more details.

Applications

What kinds of labeling tasks do you support?

We support any labeling task you imagine, from simple to complex.

For example, on the simple end, you can run image categorization projects, text categorization projects, tweet labeling projects, etc.

For example, on the complex end, you can run hate speech labeling projects with 50+ pages of guidelines, search evaluation projects, misinformation research projects, projects that require labelers with advanced CS and STEM degrees, projects to evaluate the quality of large language models, etc.