Salesforce Unveils xGen-Sales and xLAM AI Models

Salesforce (NYSE: CRM) has unveiled two recent AI models, xGen-Sales and xLAM, designed to revolutionize sales and operational efficiency. These innovative models are set to transform how businesses deploy autonomous AI agents, enhancing scalability and performance.

xGen-Sales, a proprietary AI model developed by Salesforce, is designed to automate complex sales tasks through its integration with Agentforce. This advanced model excels in industry-specific functions, providing rapid and accurate responses. It automates critical sales processes such as generating customer insights, enriching contact lists, summarizing calls, and tracking pipelines. Salesforce’s evaluations show that xGen-Sales surpasses even larger models in performance, offering enhanced precision and efficiency.

Salesforce is also introducing the xLAM family of Large Action Models (LAMs), which represent the next evolution in AI technology. Unlike traditional Large Language Models (LLMs), which often require significant human intervention, LAMs are specialized in function-calling tasks, allowing them to execute capabilities within various systems and applications independently. The xLAM models include:

– xLAM-1B: Known as the “Tiny Giant,” this model features 1 billion parameters and is ideal for on-device applications, such as AI assistants on smartphones.

– xLAM-7B: Designed for academic exploration and lightweight agentic applications, this model offers quick planning and reasoning with limited GPU resources.

– xLAM-8x7B: A mixture-of-experts model suitable for industrial applications, balancing latency, resource consumption, and performance.

– xLAM-8x22B: A large mixture-of-experts model providing optimal performance for organizations with substantial computational resources.

The xLAM models deliver exceptional performance and cost-effectiveness. For example, the xLAM-1B model has outperformed larger and more expensive models and is available as a non-commercial, open-source option to advance research. Salesforce’s proprietary APIGen pipeline has been instrumental in achieving these results, with xLAM-8x22b recently ranking first in function calling on the Berkeley Leaderboards.

Rena Bhattacharyya, Chief Analyst at GlobalData, highlights the impact of open-sourcing LAM models, noting that Salesforce’s xLAM-1B makes high-performance AI accessible to a broader range of businesses. This innovation is expected to drive greater efficiency and transformation across various industries.

MaryAnn Patel, SVP of Product Management at Salesforce, emphasizes the company’s vision of augmenting sales teams with AI solutions to enhance selling efficiency and allow teams to focus on customer interactions. Salesforce AI Research continues to lead in developing cutting-edge technologies that directly impact business operations and drive AI adoption across sectors.