Salesforce AI Research
Ethical AI solutions for customer empowerment
About
Salesforce AI Research, established in 2014 and headquartered in San Francisco, California, stands at the forefront of artificial intelligence (AI) innovation, especially in the domains of generative AI and large language models (LLMs). What sets Salesforce AI Research apart is its seamless integration with Salesforce's vast reservoir of data, enabling the creation of AI solutions that are both trustworthy and responsible. This deep-rooted connection with Salesforce's ecosystem not only accelerates the deployment of new AI capabilities into the company’s core products but also ensures that these advancements have immediate real-world applications, thus benefiting Salesforce's extensive client base and impacting the broader AI landscape. A hallmark of Salesforce AI Research's work is their pioneering development of large action models (xLAMs), which are not merely focused on generating text but are engineered to understand and execute complex task sequences. This capability bridges the traditional gap between AI's theoretical potential and its practical usability, demonstrating remarkable progress in the application-oriented spectrum of AI research. By harnessing vast amounts of function and API calling data, the team meticulously refines and standardizes this data to optimize model training. This emphasis on action-oriented AI signifies Salesforce’s commitment to practical and impactful innovations in AI. The research group is also heavily involved in broadening the scope of AI functionalities which includes fine-tuning areas such as code generation and understanding, aimed at enhancing developer efficiency, retrieval-augmented generation (RAG) which improves content creation workflows, and time series intelligence that offers transformative insights derived from data trends over time. These areas highlight Salesforce’s multidimensional approach to AI, aiming to integrate these technologies deeply into agile and intelligent enterprise applications. Salesforce AI Research's commitment to ethical AI practices can be seen through initiatives like the Einstein Trust Layer. This layer is integral to Salesforce's AI Cloud, playing a pivotal role in ensuring data security by incorporating features such as dynamic grounding, zero data retention, and toxicity detection. Such measures not only safeguard AI applications but also address growing concerns regarding data privacy and bias, promoting a culture of trust and transparency. By setting high standards for ethical AI, Salesforce AI Research plays a key role in shaping industry guidelines and fostering a responsible AI environment. Additionally, Salesforce AI Research actively contributes to the open-source arena, releasing numerous projects to encourage collaboration and innovation within the global AI community. This not only accelerates advancements in AI technology but also underscores Salesforce's commitment to making AI advancements accessible and beneficial universally. Through these efforts, Salesforce AI Research is not only advancing AI technologies but also cultivating an ecosystem that values shared knowledge and collaborative progress.
