Adaptive ML, the startup enabling companies to perpetually improve their generative AI models based on users’ interactions, announced today that it has raised over €18.3 million seed round led by Index Ventures. ICONIQ Capital, Motier Ventures, Databricks Ventures, IRIS, HuggingFund by Factorial, and many prominent angel investors joined the round.
Adaptive’s vision is to empower enterprises everywhere to leverage the benefits of ever-learning generative models and deliver intuitive experiences to lift key business metrics. The company has already shipped the first version of its enterprise platform and will use the funding to continue to invest in research, and product development, and rapidly grow the team in Paris and New York.
CEO Julien Launay said “At Adaptive ML, we are streamlining the path to more intuitive, stickier genAI applications, directly driving improved user experience and business outcomes.” continuing with: “In the long term, we envision perpetually learning models absorbing every interaction to deliver experiences unique to every user.”
There is high demand for enterprises to use large language models (LLMs); however, there is currently a large gap between a generic off-the-shelf model and making it work for a specific production use case. Engaging experiences can only occur when the model is aligned with users’ goals. Getting there is a complex process that involves highly technical and specialized techniques. This complex preference-tuning process propelled OpenAI to worldwide phenomenon status, like ChatGPT. Adaptive will unlock the power of these techniques and make them accessible to enterprises everywhere. The Adaptive platform abstracts away the technical details around fine-tuning and reinforcement learning, accelerating time-to-release for LLM-based products. To accomplish this, Adaptive has an innovative approach to engineering, data, and deployment.
Bryan Offutt, Partner at Index Ventures said: “Alignment is the foundation of how users experience large language models, while pre-training teaches a model how to think, it is an alignment that teaches it how to communicate. We are excited to partner with Adaptive to bring the power of alignment to enterprises everywhere, enabling them to build production-ready models that are uniquely suited to their business.”
Adaptive is built for any enterprise interested in deploying large language models, for example in customer support, software development, and deep research efforts. For more information on Adaptive and its preferred tuning platform, please visit adaptive-ml.com.