Make sense out of your text data
Get the relevant information with NLP & Transfer Learning
We use cookies to provide the best site experience.
Ok, don't show again
Close
What we do
We bridge the gap between NLP research and the industry. With our open-source software, we enable developers to use the latest language models & transfer learning techniques for their individual tasks. Our additional features & services help enterprises to build, run and maintain production-ready NLP applications.
Contact us

Our Products
Free for developers. Full service for enterprises.




Language models
The core of modern NLP


  • Enterprise Add-ons
  • Industry-specific: LegalBERT, FinBERT ...
  • Company-specific: Tailored to your own corpus
  • Language model as a service: Regular upgrades to latest model architectures
GermanBERT




FARM
Transfer Learning for the Industry
  • Open-Source
  • Easy adaptation of language models for Document Classification, NER, Question Answering ...
  • Fast integration of custom datasets
  • Modular design of language model and prediction heads
  • Smooth upgrading to new language models
  • Fast pipelines & inference


  • Enterprise Add-ons
  • Managed Deployments
  • Integrations & Plugins
  • Model monitoring & optimization
  • Service Level Agreements
  • Prioritized Feature Requests
  • Expert Support
GitHub




Haystack
Question Answering at Scale
  • Open-Source
  • Question Answering at scale via reader & retriever
  • Readers: Transformer models like BERT, ALBERT & Co.
  • Retrievers: Fast and scalable via ElasticSearch
  • Basic labeling tool


  • Enterprise Add-ons
  • Advanced labeling tool (incl. multi-user support, quality metrics and additional input document types)
  • Continuous learning from human feedback
  • Fully managed models (we train & host, you consume the API)
  • Integrations & Plugins
  • Service Level Agreements
  • Prioritized Feature Requests
  • Expert Support
GitHub
Our Services


Workshops & Trainings

We bring you up to speed with latest developments in NLP. For executives we focus on concepts, business opportunities, market structures and integration approaches. For developers we provide introductions to deep learning, cutting-edge model architectures and recent developer tools to accelerate your work. The final content is always fitted to your needs. What distinguishes us from others: We draw from the experience of real-world NLP projects.



Custom Development & Consulting
Use Case Discovery
Through a structured format we gain an understanding of processes, information needs and data and provide impulses to identify valuable cases for NLP
Proof of Concept
We apply technologies to real customer data and build working models to assess the technological potential of NLP for the identified use case
Productization
The integration into existing IT landscapes is performed, under consideration of all relevant requirements regarding performance, robustness and data security
Operation & Maintenance
To keep our customers' focus on their business, we ensure the availability of the developed system and take care about continuous adaptions to ever changing conditions
What NLP can do for you
No magic. Just cutting-edge NLP driven by Transfer Learning.
Automatically detect and tag different types of documents.
Condense texts to their absolute essence.
Search for content not just for keywords.
Detect the key topics across large corpora of documents.
Detect & extract structured data like products, proteins or liability sums from your documents with customized entity recognition.
Automatically reply to frequently occurring questions.
We love diversity and enjoy working with clients of all sizes:
From small tech-startups to large international corporates.
Clients & Partners
What our clients say

Management
As experts from a relatively young industry, we know about the potential of machine learning but see the untapped opportunities at the same time. Bridging this gap unlocks enormous value but also requires thoughtful and responsible change. Therefore, bridging this gap is not just a matter of technology or its implementation - it is also a matter of how change is approached. Introducing methods of NLP, as one of the most disruptive disciplines of this century, will make machines understand - but also people need to do so. Bringing all together - approach, technology and especially our clients - is the recipe for generating value. This is the rationale behind deepset and it is what drives our work everyday.
Milos Rusic
Co-Founder
Milos introduced methods of machine learning to large companies and mature industries. Having a sound understanding of the technologies he believes in the value of applying them to all kinds of business problems. Milos gained experience in research, consulting and AI start-ups.
Malte Pietsch
Co-Founder
Malte created Machine Learning models in various sectors with a focus on NLP. Fascinated by the speed of recent developments in deep learning he enjoys bridging the gap from research to business applications. Previous stops: plista, TU Munich and Carnegie Mellon University.
Timo Möller
Co-Founder
Timo is the tech guru at deepset, he has been researching Neural Networks since they still had one hidden layer and coded a deep object recognition system from scratch. He has been working in the industry to apply machine learning models at all scales and believes it is important to democratize AI knowledge through open sourcing and education.
Management
As experts from a relatively young industry, we know about the potential of machine learning but see the untapped opportunities at the same time. Bridging this gap unlocks enormous value but also requires thoughtful and responsible change. Therefore, bridging this gap is not just a matter of technology or its implementation - it is also a matter of how change is approached. Introducing methods of NLP, as one of the most disruptive disciplines of this century, will make machines understand - but also people need to do so. Bringing all together - approach, technology and especially our clients - is the recipe for generating value. This is the rationale behind deepset and it is what drives our work everyday.
Milos Rusic
Co-Founder
Milos introduced methods of machine learning to large companies and mature industries. Having a sound understanding of the technologies he believes in the value of applying them to all kinds of business problems. Milos gained experience in research, consulting and AI start-ups.
Malte Pietsch
Co-Founder
Malte created Machine Learning models in various sectors with a focus on NLP. Fascinated by the speed of recent developments in deep learning he enjoys bridging the gap from research to business applications. Previous stops: plista, TU Munich and Carnegie Mellon University.
Timo Möller
Co-Founder
Timo is the tech guru at deepset, he has been researching Neural Networks since they still had one hidden layer and coded a deep object recognition system from scratch. He has been working in the industry to apply machine learning models at all scales and believes it is important to democratize AI knowledge through open sourcing and education.

Resources
FEEL FREE TO CONTACT US
engage@deepset.ai
Write Close
Close
Get in touch
By clicking the button you agree to our Privacy Policy