Boltzbit, an innovative generative AI service, has raised £1.6 million in its seed round, led by Speedinvest and joined by IQ Capital, to transform its cutting-edge AI tools into an enterprise-ready platform. Boltzbit builds on its founders’ years of research and direct experience in working with AI and machine learning, helping them create a system that automatically avoids the problems that have plagued previous implementations of AI while still being accessible to almost everyone.
The London-based startup promises to transform and democratise AI, by reducing the resources needed for people to use it, opening it up to new users and sectors, and helping to unleash a new wave of AI-driven solutions to human problems.
Improving an imperfect technology
Despite its potential, AI has not always had a good press. There have been many examples of AI’s mistakes, for example when training to diagnose illness, using the institution name, rather than actual scans as an indicator of disease. Other examples have been more pernicious, with AIs failing to recognise bias in the data and, effectively, becoming prejudiced themselves but lending an air of impartiality to the machine’s flawed suggestions.
But AI could also be inaccessible. The resources required, from the data itself, to the machines needed to analyse it, and the data scientists to manage the whole process limited the application of AI. An issue noted by lead investor Speedinvest. “One of the tricky parts in A.I. is to develop models and algorithms that can understand and analyse troves of data, and generative models are one of the most exciting approaches to achieve this goal,” said Ricky Hao, a partner at Speedinvest. “Boltzbit has developed unique techniques to overcome these challenges, and a low code platform to allow enterprises to easily train and deploy deep generative models.”
Boltzbit’s founders came to the problem with a wealth of experience. Co-founder and CEO Dr Yichuan Zhang first met fellow co-founder and chief research scientist Dr Jinli Hu while they were working on their PhDs together. He met the third co-founder and chief technology officer, Dr Hanchen Xiong, when they worked together at Zalando in Germany. They also brought a mix of academic and commercial knowledge to form Boltzbit. Together, they aim to disrupt the status quo with a platform that is accessible to all, allowing them to run thousands of AI models on their platform, and all without coding.
A different approach to solving AI’s problems
Boltzbit aims to harness the power of generative AI. This is more resource-intensive, since it can predict whole models, rather than single labels but the added complexity helps to produce safeguards against the AI developing biases. “Generative AI is more resilient to biases,” Zhang told TFN. “It is trained to capture all perspectives of data, so it is much harder to bias.”
That use of data also helps to power the low-code interface that Boltzbit are developing for enterprise users. Users may only need to complete something as simple as an upload. “Our platform only needs the user to upload a dataset to create an AI as a ready-to-use cloud-based software solution,” said Zhang. “In particular, the low-code feature means the human intervention in the data-to-AI development on our platform is completely optional. This reduces the AI development time from days to hours.”
The simple-to-use model means that almost anyone will be able to use Boltzbit, the main requirement will be having enough data for the platform to create models. In testing, the team even found it was able to correct data! “We have seen generative AI can make accurate predictions in asset price and investment risks,” said Zhang. “This is because our generative AI can learn to correct potentially inaccurate information in its input, which is impossible for many other non-generative AIs.”
Developing an AI solution for enterprise
Having developed their model, Boltzbit plan to use the funding to bring generative AI to the market. The funds will help them scale up, open a Berlin office and grow their engineering and research capacity, helping to bring the latest research into a practical solution for financial services, healthcare, and search engine optimisation.
They will also be boosting their marketing activity, recognising that previously AI had been seen as only affordable by large corporates. “Boltzbit aims to be the leader of this exciting trend to create the next generation of AI to help all businesses to solve challenging problems using the power of data,” says Zhang. Instead of being limited to those with deep pockets, Zhang thinks AI will become commonplace, “we believe generative AI will be an interactive part of everyday life and work.”