Why we built Bedrock
Today marks the official launch of Bedrock, the culmination of 18 months of developing a platform which encapsulates how we believe artificial intelligence solutions should be deployed within enterprises - rapidly and responsibly.
The back story.
When Linus, Sil and I co-founded BasisAI almost two years ago, we saw first hand how there was a widening gap between the AI capabilities of technology giants and the rest of the world. Silvanus and Linus had seen how data science and machine learning were deployed at internet-scale at Twitter and Uber, and I had witnessed the opportunity, but also the bottlenecks, faced within more traditional enterprises when driving Smart Nation initiatives for the Singapore government.
We knew that the opportunity for machine learning adoption outside of the big tech companies was monumental. From the public sector, to financial services and telcos, many organisations wanting to be more data-driven and automated in their approaches were facing (and continue to face), a whole host of challenges in getting machine learning (ML) to work for their businesses. Some of these challenges involved finding the right business-tech translators to cut through the hype in order to find the kernel of value from ML. Others included finding the right combination of human talent and technology to enable data science and DevOps teams to be productive.
But enterprises, however ambitious and committed to digital transformation, work very differently from startups and data-native companies. Many of them are highly regulated and work in sectors where trust between them and their customers has been hard won over the course of many years. Their demands from artificial intelligence (AI) are not just for it to be deployed quickly to generate exponential business upside, they want to be able to trust their AI systems - to make them transparent, accountable and fair. We have always believed that speed and governance of AI systems aren’t diametrically opposed, if you build ML systems correctly from the ground up, you can achieve both. And it sets the foundation both for innovation and trust.
We decided to base the company out of Singapore because we felt there was a tremendous pool of talent as well as a progressive set of companies who operate out of the region. Singapore has great ambitions to be a smart city and to harness technology to transform its businesses and make lives better for citizens. It has coupled an open economy, dynamic economy with a strong system of governance which has given it a reputation as a trusted hub to do business in. It is no coincidence that the philosophy of our product mirrors the place we chose to headquarter the company.
What’s fascinating is the multidisciplinary nature of conversations we’ve had and the overlapping disciplines that exist. On one hand we have seen how data scientists and software engineers have different methodology and mental models when approaching their work necessitating different tool sets. We also see the challenges of productionising machine learning (often referred to as MLOps amongst engineering circles). On the other hand, we hear regulatory and ethical concerns around black box artificial intelligence systems. When starting out with BasisAI, we asked ourselves if the answer to both these issues lie in technology that addresses both these concerns together, or could be inspired by approaches from one paradigm, rather than being in opposition. And that was the driving force behind starting to develop Bedrock.
Bedrock - the foundation for responsible AI.
Over the past year, we’ve been focused on building a platform that helps data-driven enterprises achieve a faster time to impact from machine learning, whilst providing the governance-by-design that is required for developing trustworthy and performant AI systems. Bedrock is an end-to-end machine learning platform which empowers data-driven enterprises to deploy AI in the real world responsibly. It is the command centre that sits at the heart of the entire ML development, deployment, and continuous learning process. We see two crucial benefits that Bedrock provides for our customers:
- Faster time-to-impact for machine learning in your enterprise
Bedrock enables MLOps practices which reduces the time-to-market of machine learning systems by up to 70%. It does so by automating workflows for training, reviewing and deploying machine learning systems in a reproducible manner at scale. It allows technology leaders to leverage the best of their technical teams by enabling a single coordination point for collaboration, and removes the frustration from data science teams, allowing them to focus on what they do best.
- Achieving AI governance-by-design
Having a strong AI governance platform enables machine learning adoption to be accelerated and used responsibly from the ground up. Every machine learning engine built on Bedrock is more reproducible and explainable. A single-pane-of-glass enables digital auditability and the ability to trace the provenance of models. Bedrock is also designed to detect unintended bias in complex models on attributes the organisation has deemed ethically objectionable. Constraints can then be placed on the models to mitigate bias based on determinants like gender or race. Only trusted and unbiased AI models make it through to production, and they continue to be monitored after go-live, alerting the technology teams before things go wrong.
Bedrock can be leveraged by enterprise businesses in a few ways. We either can work with you from conception to execution to identify high impact AI opportunities, develop bespoke machine learning models which we then deploy and sustain for you with Bedrock. If your organisation has data scientists who are already developing their own models, Bedrock can be easily integrated for use by your data and IT operations teams to give them control but enable AI governance by design.
Building trust in AI as adoption grows.
Disruptive technology will always move ahead of what society is comfortable with. That’s what makes it powerful. But for it to become mainstream, technology needs to be trusted by the people who use it, and the people who often are unwitting consumers of it. We believe that artificial intelligence is at this inflexion point - where what’s slowing more widespread adoption is trust. It’s an imperative that BasisAI is here to solve, not just for the good of our customers, but for the society and future generations for whom AI will be part of the technological and social fabric.
Thank you for reading our story.