About Antigranular
The world is full of data. Data which helps us draw insights, develop innovative solutions, and drive positive social change. There is a problem though. Less than 1% of this powerful data gets utilised. The rest? It sits in silos across organisations and is locked tightly due to privacy concerns.
Privacy-enhancing technologies (PETs) hold the key to unlocking the potential of the world's most impactful data. PETs represent a range of methods, systems, and tools that ensure the secure handling of sensitive data. They facilitate data processing tasks without compromising individual privacy.
While there's plenty of theoretical work focused on PETs, their practical implementation in data science often falls short. Antigranular was established to bridge this gap. Our mission is to offer a practical way for data scientists to utilise PETs and encourage their broad adoption.
Our implementation of a differential private framework for data scientists is designed for optimal simplicity and efficiency. Data scientists interact with sensitive data through Jupyter notebooks using the %%ag magic command. This command facilitates the execution of Python code in a remote, secure environment, we call it “private-python”. It is a restricted variant of Python that limits operational capabilities and ensures that data frames adhere to approved differential privacy mechanisms.
Antigranular aims to cultivate a community where privacy-conscious data scientists can come together to learn, compete in hackathons, develop, and enhance these technologies. It encourages experimentation, refinement, and collaboration to enable the widespread adoption of PETs.
AGENT
Antigranular Enterprise (AGENT) is a data privacy platform built specifically for enterprises to facilitate interactive data science with sensitive data sources. It builds upon the standard features of Antigranular by providing self-hosting capabilities, allowing organisations to deploy AGENT on their own infrastructure and configure it to support their preferred libraries, privacy budgets and authorisation flows.
Like Antigranular, AGENT enforces the requirements of differential privacy by ensuring that all queries and processes comply with strict privacy standards, thereby ensuring a secure data environment.
AGENT is built to manage and track the privacy budget for each user and team, associated with every sensitive data source. This prevents breaches of pre-established privacy constraints. The system is built upon custom Jupyter Kernels, facilitating the execution of client code within a secure, isolated environment, 'private-python'.
For data scientists, AGENT offers a seamless and flexible coding experience. They can effortlessly switch between 'private-python' for remote execution and standard Python for local execution, using simple Jupyter notebook magic commands.
All users of the AGENT interact with the system through the AGENT Console, a user portal that allows administrators, data owners, and data scientists to manage all aspects of the system effectively. AGENT is designed around core principles like zero-trust security, self-contained infrastructure, minimal support needs, detailed user control, cost efficiency, and an emphasis on ease of use
With AGENT you can connect to over 200 data sources and drive meaningful insights in your organisation while keeping granular data private and confidential.
About Oblivious
Antigranular, AGENT, and OBLV Deploy are cutting-edge privacy-by-design products developed by Oblivious. Oblivious, privacy and confidentiality company specialising in PET solutions for businesses, was established in 2020 by Robert Pisarczyk and Jack Fitzsimons and is based in Dublin.
Through Oblivious solutions, enterprises can adopt a zero-trust architecture and securely deploy cloud services within a trusted execution environment
As a part of our commitment to the PETs community, Oblivious organises the annual Eyes-Off Data Summit. The event offers a forum that unites diverse stakeholders where they can collaborate on advancing PETs and privacy.
With a strong focus on nurturing innovation, interaction, and knowledge sharing, the event seeks to shift the trust paradigm from traditional, document-based methods to more robust technological assurances, work towards championing responsible innovation and advocating secure data sharing practices.