Salesforce unveils new AI models xGen-Sales and xLAM to enhance business management. Among these, the xGen-Sales model is designed to automate sales processes within Agentforce. This proprietary model is trained to take over routine sales tasks and help human agents focus on more strategic work.

In addition to xGen-Sales, Salesforce has unveiled xLAM, a new family of Large Action Models. These models are engineered to tackle more complex tasks and generate actionable insights that drive business operations forward.

These AI models represent a significant leap in Salesforce’s AI capabilities. They are designed to be easily integrated into existing systems developed by Salesforce AI Research.

In the blog, let’s learn about Salesforce’s new AI models: xGen-Sales and xLAM. Here we go!!

Salesforce New AI Models: xGen-Sales And xLAM

xGen-Sales is part of an emerging category of AI models known as Large Action Models (LAMs). Traditional AI models, like Large Language Models (LLMs), are primarily focused on generating content, such as text or images. However, they often require human oversight and input to function effectively.

LAMs are designed to go beyond just creating content, they can actively trigger and execute actions within other systems and applications. This capability allows AI agents to perform tasks independently, automating processes that previously required human intervention.

Salesforce’s xGen-Sales model is a prime example of this new technology, designed specifically to automate sales tasks within Agentforce. It’s a powerful tool that can take over routine sales activities, enabling businesses to operate more efficiently.

Salesforce AI Research has introduced a broader family of LAMs known as xLAM. These models are particularly noteworthy because they offer several advantages over existing AI models. They are more cost-effective, operate faster, and deliver higher accuracy.

The xLAM-1B model is open-source and non-commercial, making it available to the research community to help advance AI science. Meanwhile, Salesforce uses a more advanced version of this model for its Agentforce system, ensuring that its customers benefit from the highest levels of performance and efficiency.

How xGen-Sales Improves Sales Automation and Efficiency

Salesforce has fine-tuned the xGen-Sales model to perform specific industry tasks accurately. For example, the model can generate detailed customer insights, enhance contact lists, summarize sales calls, and monitor the sales pipeline.

By integrating xGen-Sales into Agentforce, Salesforce has significantly boosted the effectiveness of sales agents. These AI-powered agents can now autonomously manage the sales pipeline and offer coaching to sales representatives with a higher degree of accuracy and efficiency.

Salesforce’s evaluations have shown that xGen-Sales outperforms many larger models in speed and accuracy. This demonstrates the model’s power and effectiveness, making it a standout tool in sales automation.

How Salesforce Trained xLAM Models with APIGen

To develop the xLAM models, Salesforce AI Research built a powerful tool called APIGen, which generates high-quality synthetic data. This tool was key to the rapid success of the xLAM models.

The xLAM Family: Four Powerful Models: Salesforce’s xLAM family consists of four distinct models, designed for different use cases.

  • Tiny (xLAM-1B): The “Tiny Giant”Overview: This model features 1 billion parameters and is perfect for on-device applications where larger models would be impractical.

    Use Case: It can be used to create powerful AI assistants that run directly on devices like smartphones or tablets, making them responsive and efficient even with limited computing power.

  • Small (xLAM-7B): The Academic ExplorerOverview: With 7 billion parameters, this model is designed for quick academic research, particularly in environments with limited GPU resources.

    Use Case: It’s ideal for performing tasks like planning and reasoning in lightweight environments, making it a great choice for researchers or small-scale applications.

  • Medium (xLAM-8x7B): The Industrial WorkhorseOverview: This 8x7B mixture-of-experts model strikes a balance between latency, resource consumption, and performance.

    Use Case: It’s well-suited for industrial applications that require speed, efficiency, and strong performance without overwhelming computational resources.

  • Large (xLAM-8x22B): The Performance OptimizerOverview: The largest in the xLAM family, this 8x22B mixture-of-experts model is designed for organizations with significant computational resources.

    Use Case: It offers optimal performance for demanding applications, making it ideal for businesses that need to maximize the capabilities of their AI systems.

Models with varying sizes and capabilities, provide flexibility for different applications, allowing businesses and researchers to choose the best fit for their needs. The rapid success and performance of the xLAM models, thanks to the innovative APIGen pipeline, highlight Salesforce’s leadership in advancing AI technology.

Availability:

  • An open-source version of the xLAM suite is available on Hugging Face for community review and benchmarking.
  • xGen-Sales has completed its pilot and will be generally available soon.

Wrap-Up

Salesforce’s recent announcement of their next-generation AI models, particularly the xGen-Sales and xLAM family of models is a significant leap forward in the realm of sales automation and customer relationship management (CRM).

As we continue to explore these advancements at The Pinq Clouds, we’re excited about the potential they hold for our clients. Whether it’s enhancing sales processes or enabling smarter customer interactions, these AI innovations redefine the future of sales and service in the Salesforce ecosystem.

Stay tuned for more insightful blogs and reach out to our AI- expertise for more information on new AI models xGen-Sales and xLAM.