NEW YORK, NY, Nov. 9, 2023 – Merlyn Mind, an AI company focused on building domain-specific AI platforms and solutions for improving human productivity in enterprises, has announced a strategic partnership with Google, including the adoption of Google Cloud to power its AI infrastructure needs. By adopting Google Cloud, Merlyn Mind’s training of large language models and development of domain-specific AI will be faster, safer, more sustainable, and more cost-effective.
Merlyn Mind’s scientists, engineers, and innovators are working at the leading edge of AI to develop large language models (LLMs) and platforms that plan and reason, have long-term memory, have better models of human cognition, have a better understanding of the world, and perform actions on behalf of their users.
Different domains pose different challenges due to the unique workflows, constraints, and requirements of people in different roles.In education, for example, teachers, students, and administrators have different needs, all of which custom AI assistants can address to improve productivity and enhance outcomes. Likewise, in health care, therapists, doctors, and patients require different approaches to automation and knowledge management.
Merlyn Mind’s approach to tackling domain-specific problems involves building domain-specific AI platforms powered by small, efficient, and highly performant LLMs which are fine-tuned for specific use cases. The LLMs in the platform will operate on enterprise-specific data and content and are trained to minimize hallucinations, which is critical to enterprise use. This platform enables the creation of multimodal AI assistants that are deeply integrated into human workflows to improve productivity by automating aspects of these human workflows. Merlyn Mind works with enterprises in a consultative approach to define opportunities for AI to deeply impact enterprise productivity and either develops or providesAPIs to enable the development of highly efficient, cost effective, safe and impactful solutions based on Merlyn Mind’s generative AI platform for the domain. Finally, an important aspect of Merlyn Mind’s approach is their focus on preserving data privacy for enterprises by ensuring that enterprise data is not compromised.
Merlyn Mind does not use enterprise data to train their models and is compliant with all relevant privacy regulations. Further, Merlyn Mind’s models and assistants are built to operate in any environment that enterprises prefer.
Education, Merlyn Mind’s inaugural domain, is a great example of this approach. The company’s generative AI platform has a suite of small and efficient LLMs fine-tuned for the education domain. These enable hallucination-resistant, efficient, private, and safe operation on top of content and data that school districts use. Their first assistant Merlyn, built on top of this platform, is an AI voice assistant focused on improving teacher productivity and enhancing learning outcomes. The Merlyn voice enabled assistant enables tool control and automation of teacher workflows on the web and delivers the power of generative AI into classrooms everywhere.The Merlyn assistant is compliant with regulations that govern privacy in the domain such as COPPA, FERPA and GDPR, and Merlyn Mind ensures that no sensitive data such as PII (Personal Identifiable Information) is used to train their models and that voice data is deleted after processing.
“The most meaningful solutions in a vertical domain will come when teams develop AI in a purpose-built way,” says Dr. Satya Nitta, Merlyn Mind’s CEO and co-founder. “These platforms and assistants must be imbued with a deep awareness of domain-specific workflows and needs and will understand specific contexts as well as domain-specific data. Google’s cloud computing platform will help us meet these conditions and utterly transform industries, ushering in untold gains in productivity and enabling humans to reach their highest potential.”
As a domain-specific AI company, it is important for Merlyn Mind that the infrastructure be performant and cost-efficient to run these AI workloads. With Google’s highly scalable AI infrastructure – including v4 of its custom-designed Tensor Processing Units, which are 1.2x-1.7x faster and use 1.3x-1.9x less power than the competitors – it was a natural choice for Merlyn Mind to make Google Cloud’s platform its playground for experimentation and production deployments.
Merlyn Mind’s deep understanding of how to deploy AI in a domain specific fashion and Google’s highly scalable cloud infrastructure will result in a highly synergistic partnership. We both believe this partnership will accelerate solutions to some of the most challenging problems in specialized domains like education and healthcare with the help of rapidly evolving AI advancements.
To learn more about Google Cloud for EdTech’s GenAI solutions and partnership programs, go to https://goo.gle/edtech.