Your Success is Our North Star

Addy is a fully configurable document analysis platform powered by machine learning. Though Addy requires no coding experience, we work closely with our customers to ensure their business needs are met. We will walk you through the training process and work with your stakeholders through the final testing and training of the ML models.


Once you become familiar with how Addy works, you may choose to build your own models. The system is highly configurable and customizable to your precise needs and will become smarter the more you use it.

1. Select Your Data Sources

  • Can upload your own documents

  • Can web search for relevant documents through the Addy interface

  • Can select one of the data sources in our library 

2. Train Your Models

  • Addy ingests your data and parses the documents into individual sentences

  • Select the sentences, keywords, and entities that represent the type of information you want to find by giving each a thumbs up

  • Identify sentences that do not comply with your search (such as false positives) by giving them a thumbs down

  • When you find examples of sentences that match what you are looking for, tag them as “gold sentences.” Addy will learn and leverage this information to find more semantically similar sentences.

Over time, Addy will learn and improve your models as you use them. After you have trained the system with approximately ten thumbs up, ten thumbs down, and a few gold sentences, you are ready to start testing the model on documents to determine its accuracy. You will continue to make improvements as needed to refine the results.

3. Prebuilt Models

Templated models are prebuilt into Addy so you can hit the ground running. We’ve designed prebuilt models purpose-built for the healthcare industry, finance, legal, and HR, each a fully-formed ML system that can be used immediately or customized to suit your specific needs.


Models are deployed through a user-friendly interface called Document Review Manager. The interface is fully customizable to your workflows and can be deployed quickly to accelerate value.

The user can view and analyze document results from the interface to determine next steps. If the model flags a sentence, the user can easily click through to view it within the context of the document to determine relevance.

From the Addy dashboard, the user can then assign an action within the document workflow, such as sending an approval to a scheduler to make an appointment (in the case of patient referrals).

 The document review workflow is fully automated within the Document Review Manager, saving time, improving accuracy, and increasing job satisfaction and retention of valued employees.