Hi, we’re Bitfount 👋
Bitfount is a federated analytics and machine learning platform that makes extracting value from sensitive data easy, fast, private, and secure. For data custodians and data scientists or researchers partnering to achieve better insights from data, Bitfount combines the best of data collaboration design, with advanced privacy-preserving capabilities, while playing nicely with all of your existing tools and crucially not requiring the transfer of any raw data.
We set out to build Bitfount guided by one core belief:
Society’s data superpowers are gathering dust due to a problem which is easy to describe but hard to solve: The data required to solve the world's most pressing problems isn't the kind that's easily shared. Tighter privacy regulations and the limitations of current data infrastructure have created painful, slow, expensive and often ineffective workarounds for what should be seamless collaborations between trusted partners.
Data collaboration today is a painful, messy business.
As anyone who has attempted to set up a collaboration around sensitive data will know, the current process is generally very painful and slow. Valuable datasets languish in silos as a result of regulatory or commercial sensitivity concerns, incompatible data management solutions, lengthy contractual processes, or just plain lack of understanding of what data is available for which purposes within an organisation. Even if these hurdles can be overcome, data custodians do not always have access to the data science talent or data partnerships required to extract enough value from the data they gather every day.
As a result the world is being deprived of the transformative impact this data could have when properly harnessed. Everything from better therapeutic treatments for our patients, to fairer credit scoring for our neighbours, to more efficient energy distribution systems, to safer streets for our children, the list goes on and on. While different industries will have different constraints or goals, the fundamental problem across all of these missed opportunities remains the same: Somebody has the right data to solve a problem, but can’t make full use of it by themselves. Somebody else knows how to use it, but can’t safely or easily get access. Bitfount is the solution for both of these sides.
But it doesn't have to be this way.
Bitfount’s goal is to make the world’s intractable data interactable. To do that, we’re building a missing piece of global infrastructure: a secure network for sending algorithms to data instead of the other way around. From SQL queries to training and deploying of deep neural networks and other machine learning models, and from single datasets to entire data consortia, we want to help build a world in which organisations can easily share the benefits of their data with each other. Where collaboration between data custodians, data scientists, and governance teams occurs seamlessly, without compromising on data privacy or quality of analysis.
The future of data collaboration does not look like its past.
The value of data is inseparable from how it is used, and therefore it is inevitable that access rights to data will be inextricably linked to the algorithms someone is allowed to run on that data. This is what Bitfount enables, while leaving raw data in its original location, safe behind the data custodian’s firewall.
We’ve designed Bitfount to harness the best parts of open source and Software-as-a-Service (we call it “OpenSaaS”): The transparency and extensibility of an open source developer SDK, with the user experience, availability/uptime and rich functionality of a beautifully designed SaaS product. This approach, together with some important architectural decisions which we’ll be writing more about soon, make Bitfount much easier, faster and cheaper for IT teams to deploy than other federated solutions.
The platform already includes a bunch of powerful functionality like flexible and granular usage-based access controls (more on that here), a catalogue for dataset discovery, full audit history, support for custom models, algorithms and data sources, cryptographic tools such as differential privacy and secure multi-party computation (SMPC) for additional privacy guarantees, workflow integrations like webhooks and loggers for your favourite ML experimentation platforms, and more.
We're just getting started.
In our previous roles, almost everyone on the Bitfount team has experienced the pain of getting access to the right data at the right time, whether for training some of the most widely-used machine-learning models on the planet, for carrying out cutting-edge research at some of the world’s leading universities, or for powering commercial collaborations between companies. We’re passionate about removing this huge source of friction for others.
We’re extremely excited about the talented team we’ve brought together, and we’re honoured to be joined by some great investors in pursuing this vision (read our funding announcement here). If you’re interested in learning more about what we’re up to, visit us at bitfount.com to join our waitlist or book a demo. We’re looking forward to making the world’s intractable data interactable, together!