• The companies will develop innovative Japanese LLM for private cloud usage through Fujitsu Kozuchi AI services 

Kawasaki, Toronto and San Francisco – 

Today, Fujitsu announced a strategic partnership with Cohere Inc., a security and data privacy-focused enterprise AI company headquartered in Toronto and San Francisco. The strategic partnership will focus on developing and providing large language model (LLM) that enable enterprises to leverage industry-leading Japanese language capabilities that deliver improved experiences for customers and employees.

In addition, Fujitsu has made a significant investment and entered into a strategic partnership between the two companies. As part of the partnership, Fujitsu will become the exclusive provider of jointly developed services on the global market. Fujitsu plans to provide the jointly developed AI technology to customers through Fujitsu Data Intelligence PaaS, a cloud-based all-in-one operation platform, and Fujitsu Uvance, a cross-industry business model to solve social issues.

Additionally, the two companies will jointly develop Takane (tentative name), an advanced Japanese language model based on Cohere’s frontier enterprise-grade LLM. Fujitsu plans to start providing the AI model through Fujitsu Kozuchi, starting in September 2024. Takane (tentative name) will be offered for private environments, such as private clouds, to provide the best combination of outstanding AI services in a guaranteed secure environment for enterprise data.

Takane (tentative name) is based on Cohere's latest LLM, Command R+, which features enhanced retrieval-augmented generation (RAG) ⁽¹⁾ capabilities to mitigate hallucinations. It is a multilingual model trained on proprietary data from scratch, ensuring safety and transparency. Takane (tentative name) leverages Fujitsu's expertise in Japanese language training and fine-tuning technologies, and Cohere's enterprise-specific technologies.

Takane (tentative name) will focus on the critical needs of specific industries and businesses to boost productivity and efficiency. These models will be developed as a service that can be utilized in private clouds for customers that require high security, such as financial institutions, government agencies, and R&D units.

In addition to its high performing generative AI models, Cohere also has best-in-class Embed ⁽²⁾ and Rerank ⁽³⁾ models to provide advanced enterprise search applications and RAG technology. These solutions enable companies to unlock real business value from their data.

Fujitsu has years of experience with R&D of knowledge graphs, a knowledge processing technology, and has developed a knowledge graph extended RAG technology that converts a company’s diverse and large-scale data into knowledge graphs that LLMs can reference, and generative AI auditing technology that achieves generative AI that is in compliance with corporate and legal regulations. Fujitsu plans to release a knowledge graph extended RAG technology from Fujitsu Kozuchi in July 2024, and generative AI auditing technology in September 2024.

In addition, Fujitsu’s generative AI amalgamation technology, which it plans to offer from Fujitsu Kozuchi starting in August 2024, will be combined as a part of Takane (tentative name) models developed through this partnership and various models for specific domains and existing machine learning models. This model will power generative AI capabilities designed to meet the high standards required of enterprises.

Through joint development, Fujitsu and Cohere will further promote the utilization of AI by companies, and accelerate digital transformation across global markets.

Vivek Mahajan, Corporate Vice President, CTO, and CPO, Fujitsu Limited, commented: “We are very pleased to strengthen our generative AI for enterprises portfolio through this partnership with Cohere. Fujitsu has developed a knowledge graph extended RAG technology for logical inferences and a generative AI amalgamation technology for automatic generation of specialized generative AI models to meet the diverse needs of companies. Combining these with Cohere's latest highly secure enterprise LLMs, we aim to provide businesses with powerful and adaptable AI solutions that address specific needs and accelerate the adoption of generative AI globally.”

Aidan Gomez, Co-founder and CEO, Cohere Inc., commented: “We believe that this strategic partnership with Fujitsu is a truly important step in offering world-class LLM capabilities to one of the most important enterprise markets in the world. For AI technologies to reach their full potential, we need to be able to meet enterprises where they are, whether that means in their own cloud environment, or in the languages that they do business. We are incredibly excited that our work with Fujitsu will help to unlock the enormous potential of Cohere’s technology to power the next generation of Japanese businesses.”

• [1] RAG:
  A mechanism through which the LLM acquires and utilizes knowledge outside of training, also referred to as retrieval augmented generation.

• [2] Embed :
  Embed is a model that converts items input into it, such as text, into vector representations. It can be used when checking whether or not texts are similar to each other.

• [3] Rerank :
  A model that allows for the close examination and highly accurate re-ranking of information obtained from a simple similarity search when the LLM is given a reference document.