Prodigy
Berlin-built, self-hosted annotation tool from the makers of spaCy, sold as a lifetime licence
Quick Overview
| Company | ExplosionAI GmbH |
|---|---|
| Category | Data Labeling Platforms |
| Headquarters | Berlin, Germany |
| Founded | Not stated |
| EU Presence | Yes - Germany (EU) |
| Data Location | Your own hardware; the software never connects to the vendor |
| Open Source | No |
| GDPR Compliant | Yes |
| Self-Hosting | Yes, self-hosted by design |
| Main Features | 23 annotation interfaces, text, image, audio and video annotation, active learning for NER and text classification, Python recipes, SQLite, MySQL and PostgreSQL storage, SSO with OpenID Connect (company licence) |
| Pricing | $390 personal lifetime licence; $490 per seat in packs of 5 for companies (excl. tax) |
| Best For | Data scientists who annotate their own NLP and vision data in code |
| Replaces | Scale AI, Labelbox |
Detailed Review
Prodigy comes from ExplosionAI, the Berlin company behind spaCy. It is a Python-installed web application that you run on your own machine or server, with annotation workflows written as scripts called recipes.
The software licence is concluded with ExplosionAI GmbH, registered at the Amtsgericht Berlin-Charlottenburg under HRB 181081 B, and the site states the software never phones home.
Annotation as code
Recipes cover named entity recognition, span categorization, text classification, dependency parsing, coreference, image, audio and video annotation and transcription. The built-in active learning models for NER, text classification and parsing suggest the examples worth labeling next.
Data stays where you put it
Prodigy works offline on your own hardware, stores annotations in SQLite, MySQL or PostgreSQL, and exports JSON. The vendor states that no data ever needs to leave your servers.
Licence rather than subscription
A personal licence costs $390 and a company seat $490 in packs of five, each with 12 months of upgrades and a perpetual right to the versions in the order. Further upgrade years can be bought at checkout.
Limitations to Consider
Prodigy is a tool for a single technical team, not a managed service: it supplies no annotators, no workforce management and no review dashboard comparable to a full platform. Collaboration relies on your own deployment and the company licence adds SSO but not hosted workspaces. Prices are in US dollars, trials are arranged by email, and the licence agreement states English law and the courts of England and Wales. For hosted team workflows see Kili Technology or Supervisely.
Who Should Use Prodigy
ML engineers and data scientists who want to script their own annotation workflows and keep every file on infrastructure they control.
Pros and Cons
Pros
- German contracting entity (ExplosionAI GmbH)
- Runs fully offline on your own hardware
- Public, one-time prices
- Open data formats and standard SQL storage
- Active learning for text tasks
Cons
- No workforce or managed labeling
- Needs Python skills to customise
- Prices quoted in US dollars
- Licence agreement under English law
Alternatives to Prodigy
Other European data labeling platforms worth comparing before you decide:
Kili Technology
France · Expert-in-the-loop labeling and AI evaluation platform
Kognic
Sweden · Annotation platform and workforce for autonomous driving data
Supervisely
Estonia · Labeling, data management and model training in one platform
Picsellia
France · Computer vision MLOps platform with built-in annotation
See all European data labeling platforms or compare them directly against Scale AI on our alternatives page.
Frequently Asked Questions
The buy page states that Prodigy runs on your own hardware and never phones home or connects to the vendor servers.
$390 for a personal lifetime licence and $490 per seat in packs of 5 for companies, excluding tax, each with 12 months of free upgrades.
Yes, within your own deployment. The company licence adds HTTP auth and SSO with OpenID Connect and seats can be transferred inside the company.