Calvin Risk — Unlocking the Full Potential of Artificial Intelligence
We announced our investment in Calvin Risk last week and wanted to take some time explaining our rationale. The company is building “Calvin”, an adaptive platform that helps companies assess and manage the technical, ethical, and regulatory risks of their AI portfolio.
At btov we believe that today, due to the availability of vast amounts of data and specialized computing hardware, AI has matured to a point where it is widely applicable. At the same time, uncertainty about the risks of deploying AI models has led enterprises to seriously underinvest in AI, which increases the risk of falling behind and being disrupted. The Calvin platform is the first universal product that is providing a solution for this problem.
Why we believe in their approach to the market
AI software revenues are approaching $100 billion per year, but at the same time a representative survey of AI leaders found that 69% believe AI risks are a serious concern and should be prioritized. In 2022 the total investment in AI risk-management systems will sum up to less than a third of what the Alphabet subsidiary company DeepMind — which is developing new AI models — is spending on its operating expenses, which stand at $1.6 billion a year. Since the investments in the space have been small, we think that at the moment there is still an extremely high marginal return.
Many of the AI algorithms being developed today are not commercially viable. There are a variety of risks inherent in the development and deployment of AI that make it challenging to transition models created in the lab to AI-enabled products that work in the real world. Algorithms are often “black boxes” (i.e. models that are difficult to interpret) trained on enormous amounts of data; if that data is systematically biased, then the resulting AI will be too. AI developers must ensure that no discriminatory relationships have been learned during the training phase of the algorithm that will result in unfair or unjust decisions. Thus, there is an acute need for evaluation metrics, context-specific assessments, and continuous monitoring to safeguard the development of AI and ensure it is compliant and works as intended.
So far, the uncertainty about the risks of deploying AI models have put many potential users and regulators on guard. Enterprises have, therefore, underinvested in AI for fear of incidents of systemic AI failures such as Microsoft’s famously abusive and racist chatbot, which had to be retracted. Increasing trust, acceptance, and understanding of the risks will become critical for the adoption of AI across industries, according to Tarek Besold, Head of Strategic AI at DEKRA DIGITAL and advisor at Calvin Risk. Understanding the risks of AI will help consumers to feel more confident in adopting new technologies, and accelerate innovation through market acceptance.
Last but not least, upcoming regulations such as the EU compliance law for AI development, make governance and risk management top of mind for corporate AI leaders. This will increase demand for products such as Calvin, which can shorten deployment cycles through efficient and trusted assessment processes while making sure a company is compliant with any new regulations.
Why we invested in Calvin Risk
As an early investor in DeepL, we saw first-hand how a smart, dedicated, and creative team based in Europe can build cutting-edge software that can gain global relevance. We believe that Calvin Risk’s product also is in a unique position. Even the most sophisticated organizations are not sure how to effectively evaluate AI algorithms across the many relevant dimensions such as performance, robustness, fairness, and explainability that are key to build trustworthy AI. On top of that, many organizations have not gathered enough experience with AI-based software products to adequately understand the risks of deploying them in the market. The Calvin platform is the first universal product that is providing a solution for this problem.
Calvin provides an overview and insights into risk levels over time and calculates the probability and potential costs of adverse incidents. The platform is useful for a variety of roles within an organization and is particularly relevant for companies in the insurance, pharmaceutical, and tech industries that have already developed and deployed AI algorithms at scale but want to improve their quality and standardization.
The founders of Calvin Risk, Syang and Julian, both have an impressive track record. Julian has a long experience as an investor at Earlybird and btov (and we love investing in our former colleagues whenever possible). Syang is an industry veteran, always having worked at the intersection of AI and risk at D-Fine and PWC. His research at ETH Zurich about applications of AI and risk management gives the Calvin platform a technological edge over potential copycat competitors.
As such, we have been closely following the development of Calvin Risk when the team was developing it as a spin-off from the ETH Zurich and have been impressed by what they’ve done so far. From their ability to close first enterprise customers, to their focus on making their platform easy to use, as well as their commitment to creating a sustainable business model, we’re looking forward to supporting them on their journey to become a global market leader.
The depth, versatility and creativity of Calvin Risk’s solution makes it pretty unique in our view. The market opportunity for a platform that manages the technical, ethical, and regulatory risks of AI is huge, and with Syang and Julian, the right technology has met the right experience. We are proud to lead Calvin Risks’s pre-seed round together with Wingman Ventures as Co-Investor. Congratulations to the entire team.
