Introduction

Discover the core principles and enterprise benefits of the AGENT platform

What's AGENT?

Antigranular Enterprise (AGENT) is a platform tailored for enterprises seeking to conduct interactive data science while safeguarding sensitive data sources.

Antigranular offers a solution for implementing differential privacy, ensuring the security of sensitive data while enabling thorough analysis. Establishing a trusted environment between data scientists and data sources, going beyond conventional binary role-based access control, allowing for nuanced and advanced privacy management.

Unlocking Potential / key benefits

AGENT enhances Antigranular's foundational capabilities by enabling self-hosting, empowering organisations to deploy AGENT within their infrastructure, and customising it to accommodate preferred libraries, privacy parameters, and authorisation workflows.

For data scientists, AGENT delivers a seamless and adaptable coding experience. Through straightforward Jupyter Notebook commands, they can effortlessly switch between a specialised version of Python explicitly designed for working with differentially private data and standard Python for local tasks, giving them the freedom and control to work in the way that suits them best.

Inside AGENT

AGENT meticulously oversees and regulates the privacy allocation for individual users and teams associated with each sensitive data source, thereby fortifying the prevention of breaches of privacy constraints. Leveraging custom Jupyter Kernels, the system ensures secure client code execution within an isolated environment known as Private Python, providing a robust shield for your data.

The AGENT Console facilitates all interactions with the platform. It is a user portal that empowers administrators, data owners, and data scientists to manage all facets of the system efficiently. With AGENT, users gain access to over 200 data sources, enabling them to derive valuable insights within their organisation while safeguarding granular data with utmost privacy and confidentiality.

AGENT’s Approach to Differential Privacy

AGENT prioritises data privacy through innovative methods, seamlessly incorporating differential privacy into data scientists' daily routines. Our core values focus on secure data handling and user-friendly practices, ensuring an optimal balance between privacy and usability.

Integration of Privacy-Enhancing Technologies (PETs) in Data Science

Antigranular leverages advanced privacy-enhancing technologies and differential privacy. These tools seamlessly integrate into data scientists' Jupyter notebooks, facilitating secure and private data analysis while maintaining individual and entity-level privacy.

Empowering Users with Customisable Privacy Parameters

We enable users to adjust the parameters of epsilon ($\epsilon$) and delta ($\delta$) in our differential privacy framework according to their specific needs. This flexible approach balances protecting individual data points and deriving insightful aggregate data analysis and inference.

Facilitating Data Sovereignty and Global Collaboration

Our strategy ensures data has passed through a data disclosure control (differentially private) before being sent to collaborators off-shore. This promotes international collaboration and allows for leveraging global data science expertise.

Minimising Data Breach Risks

We significantly reduce the risk of data breaches by controlling access to raw data and employing differential privacy. Our approach safeguards data from accidental exposure and deliberate cyber attacks, enhancing overall data security.

Ensuring Secure Integration of Multiple Data Sources

Antigranular's technology effectively combines datasets from various origins while preserving the confidentiality of each dataset. Our differential privacy measures ensure comprehensive insights without compromising the privacy of individual data sources.