Explosion logo

Explosion

Explosion is a Berlin-based software company building developer tools and tailored solutions for artificial intelligence, machine learning, and natural language processing. Its products include spaCy, Prodigy, Ellf, and Thinc.
Berlin, DE

Explosion


Project Links


Description

Developer tools for AI and natural language processing. Creators of spaCy and Prodigy.


Ellf


Project Links


Description

Ellf integrates with AI-assisted coding tools such as Claude Code and supports end-to-end NLP development, including project planning, data preparation, annotation, machine learning, rule-based pattern generation, model training, analysis, quality assurance, and evaluation. It uses spaCy, Prodigy, and the Python ecosystem, and runs through a data-private cluster hosted locally or in the user's cloud infrastructure so data, models, and custom code remain under the user's control.

Category: Developer Tooling

Prodigy


Project Links


Description

Prodigy is an extensible Python package and web application for efficient data annotation, model-assisted labeling, training, and evaluation. It supports information extraction, language model training, computer vision, audio and video annotation, prompt engineering, and custom workflows. Prodigy runs locally or on user-controlled infrastructure, keeping data and models private, and integrates with spaCy, Hugging Face, LLM providers, PyTorch, TensorFlow, and other Python tools.

Category: Developer Tooling

SpaCy


Project Links


Description

spaCy is a fast, extensible Python library for large-scale natural language processing. It provides tokenization, linguistic analysis, named entity recognition, text classification, lemmatization, entity linking, pretrained transformers and word vectors, visualizers, custom components, model packaging, and production-oriented training workflows.

Category: Developer Tooling

Thinc


Project Links


Description

Thinc is a Python deep learning library from the makers of spaCy and Prodigy. It supports interchangeable and hybrid models across PyTorch, TensorFlow, and MXNet, zero-copy array interchange, sophisticated type checking, configurable object trees, a consistent API, and lightweight installation through pip and conda.

Category: Developer Tooling