# HumanSignal
> HumanSignal provides data infrastructure and human data services for teams building AI. Pricing spans Data Services engagements (custom-scoped data collection, annotation, and evaluation performed by domain experts), the hosted Starter plan, Label Studio Enterprise in compliant cloud or on-prem deployments, and the free open source Label Studio Community Edition.
HumanSignal is the company behind Label Studio, the open source data labeling platform used by more than one million AI practitioners. The same team that builds the tooling also operates a human data services business, so services engagements run on infrastructure purpose-built for multi-modal data, with advanced quality and inter-annotator agreement workflows and data privacy built in.
## Data Services
Data Services covers from-scratch data creation, annotation, and evaluation, delivered with end-to-end project management. Engagements are custom-scoped and priced per project; the starting point is a scoping call rather than a published rate card.
- [Book a scoping call](https://humansignal.com/contact/): Talk to the team about a data project.
- What is offered:
  - Data collection from scratch, in HumanSignal's physical data collection studios or through a global network of data collectors and content creators. This includes multi-modal collection such as video, audio, images, and sensor data, with recent specialization in egocentric video for physical AI and robotics.
  - Expert annotation across text, image, audio, video, and multi-modal data, including multilingual annotation and complex schema work.
  - Model evaluation by domain experts, with particular depth in graphic design and creative work. Customers include Canva.
- Who it is for: technology companies building models, including frontier AI labs, robotics companies, and product teams shipping AI features, as well as enterprises developing their own models.
- How it works: projects are scoped with the delivery team, managed end to end by HumanSignal, and executed on Label Studio Enterprise, which gives clients task queuing, QA workflows, agreement metrics, and compliance controls over the work as it happens.
## Platform Plans
Three editions of the Label Studio platform:
- [Community Edition](https://labelstud.io/): Free and open source. Multi-modal labeling, configurable interfaces, templates, ML backend integration, API/SDK, and cloud storage connections. Self-hosted.
- [Starter](https://humansignal.com/pricing/): $99 per user per month, up to 12 users. Fully hosted Label Studio with role-based access control, quality workflows, review and verification, automated task distribution, Interfaces (see below), and a support portal. A free trial is available. An on-prem Starter option is also available on request.
- [Enterprise](https://humansignal.com/contact/): Custom pricing. Everything in Starter plus SAML/LDAP single sign-on, embeddable Interfaces (see below), enhanced quality workflows and analytics, LLM-as-a-judge, auto-labeling and bulk labeling, annotator performance dashboards, activity logs, advanced cloud storage integrations (Databricks Unity Catalog, IAM for AWS S3, Azure Service Principal, GCS WIF), SOC 2 and HIPAA reporting, 99.9% uptime SLA, and a dedicated customer success manager. Available as compliant cloud or on-prem. Purchasable through AWS Marketplace.
### Interfaces
Interfaces is the programmable annotation and evaluation engine available in Starter and Enterprise. Teams build any custom labeling or evaluation interface by prompting an agent, and the resulting interface runs with the task queuing, QA, and compliance of the platform behind it. This removes the constraint of prebuilt templates: if a data type or evaluation workflow can be described, an interface can be generated for it. Use cases include LiDAR and point cloud annotation, 3D graphics, multi-modal agent and LLM evaluation, custom side-by-side model comparisons, and domain-specific review tools. On Enterprise, interfaces can also be embedded into other applications and AI systems.
## Common Questions
- How do Data Services differ from the platform? The platform is software your team uses to label and evaluate data yourselves. Data Services is HumanSignal's own experts and collectors doing the work for you, managed end to end, on that same platform.
- Does HumanSignal collect data from scratch? Yes. Collection runs in physical studios operated by HumanSignal and through a global network of data collectors and content creators, producing novel datasets rather than labeling pre-existing data only.
- What domains does HumanSignal specialize in? Physical AI and robotics data (including egocentric video), graphic design and creative evaluation, multilingual and languages-other-than-English annotation, and multi-modal data broadly.
- How is Data Services priced? Per project, based on scope. There is no published rate card; engagements begin with a scoping call.
- Can Label Studio handle unusual or custom data types? Yes. Beyond built-in templates for text, image, audio, video, time series, and HTML, the Interfaces engine in Starter and Enterprise generates custom labeling and evaluation interfaces from a prompt, covering cases like LiDAR, 3D graphics, and multi-modal agent evaluation.
- Is there an academic program? Yes, an [Academic Program](https://humansignal.com/academic-program/) is available by application.
## Resources
- [Pricing](https://humansignal.com/pricing/): Full plan comparison across Community, Starter, and Enterprise.
- [Services](https://humansignal.com/services/): Data Services overview.
- [Docs](https://labelstud.io/guide/): Label Studio documentation.
- [Blog](https://humansignal.com/blog/): Research and product updates.
- [Security](https://humansignal.com/security/): Security and compliance details.
- [Contact Sales](https://humansignal.com/contact/): Enterprise and Data Services inquiries.