Salesforce today unveiled two new AI model families: xLAM, a new family of Large Action Models intended to perform difficult activities and produce actionable outputs, and xGen-Sales, a proprietary model trained and developed to support autonomous sales tasks with Agentforce. When combined, these models created by Salesforce AI Research will enable clients of Salesforce to easily build up and launch action-oriented autonomous AI agents, achieving previously unheard-of scale.
xGen-Sales can provide more accurate and timely responses by optimizing its accuracy for pertinent industry operations. This includes automating sales tasks like managing the sales pipeline, collecting customer insights, and enriching contact lists. With the help of this model, Agentforce sales agents can now coach representatives more quickly and accurately and manage pipelines on their own. Even Salesforce’s own assessments have shown that xGen-Sales’ capabilities have already surpassed those of other, much larger models.
An advance toward the upcoming Large Action Models (LAMs) generation of language models is xGen-Sales. LAMs (Large Language Models) are mostly utilized for content generation; in contrast, function-calling—the capacity to carry out operations within other systems and applications—is the area of expertise for LAMs. LLMs require frequent human participation. Stated differently, they have the ability to initiate the necessary actions, enabling AI agents to autonomously carry out tasks on behalf of humans. Salesforce AI Research has produced a new LAM family known as xLAM in addition to xGen-Sales. Compared to many of the larger and more complicated models that are already on the market, xLAM models provide lower costs, faster performance, and more accuracy.
For instance, using only 1 billion parameters—the variables that models learn to produce outcomes and insights—the xLAM-1B model has beaten larger and more costly models. Salesforce employs a far more effective model for Agentforce; however, xLAM-1B is a non-commercial, open-source model to assist in progressing the science with the research community.
Why it matters: AI agents that can operate as a substitute for workers are necessary for organizations in order to augment staff workloads and free up time for more strategic tasks. Agents utilizing these models will be able to discern when to assign a work to a human for completion and quality assurance because these models are aware of both their own limitations and the jobs they are meant to perform. With the latest release of its LLM Benchmark for CRM, Salesforce gives businesses the chance to compare LLMs for CRM use cases and traverse the various models available.
“Building and training your own AI models can be time-consuming, costly, and incredibly frustrating,” said Salesforce Chief Scientist Silvio Savarese. “With Agentforce, we’re able to deliver appropriately sized models, built specifically for your business with your data to drive outcomes.”
Behind the scenes: Salesforce AI Research developed APIGen, a reliable, proprietary process for producing high-quality synthetic data, in order to train the xLAM models. Good things happened almost right away. According to Salesforce’s own assessment, xLAM 8x22b ranked No. 1 for function calling on the Berkeley Leaderboards, outperforming GPT-4. The model xLAM-8x7b comes in sixth place. Both outperform models several times their own size. The xLAM family of language models consists of four models:
- Tiny (xLAM-1B): The “Tiny Giant” has 1B specifications. The model is best suited for on-device applications where larger models are more impracticable due to its small size. Using the xLAM-1B, developers may build intelligent and receptive AI assistants that operate locally on smartphones and other devices with constrained computing capacity.
- Small (xLAM-7B): With constrained GPU resources, the 7B model is intended for quick academic research. In a lightweight setting, it can be utilized to carry out reasoning and planning tasks for agents.
- Medium (xLAM-8x7B): The 8x7B mixture-of-experts model is perfect for industrial applications that aim to achieve a fair trade-off between performance, resource usage, and latency.
- large (xLAM-8x22B): The 8x22B is a huge mixture-of-experts model that enables the best possible performance for companies with a given amount of processing capacity.
The customer perspective: “The models Salesforce is delivering for its Agentforce platform are what give us confidence that we’ll have the capabilities we need to roll out strong and cost-effective autonomous AI capabilities over time,” said Rena Bhattacharyya, Chief Analyst and Practice Lead, Enterprise Technology & Services at GlobalData. “Salesforce is truly paving the way for the AI Agent revolution.”
The Salesforce perspective: “We envision a future in which sellers are augmented by AI to help them drive selling efficiency, freeing up precious time to focus on their customers,” said MaryAnn Patel, SVP, Product Management at Salesforce. “The xGen-Sales model is purpose built to help companies build generative AI solutions that will augment the work of their sales teams with Agentforce.”
Salesforce’s AI research lab, known as Salesforce AI Research, creates new technological innovations in the industry. The group, which consists of engineers, product managers, and researchers, is striving to directly influence product development through fundamental research, thereby shaping the future of AI for businesses.
Also read: Unveiling the Ethical Imperatives: Navigating the Intersection of AI and Cybersecurity
Do Follow: CIO News LinkedIn Account | CIO News Facebook | CIO News Youtube | CIO News Twitter
About us:
CIO News is the premier platform dedicated to delivering the latest news, updates, and insights from the CIO industry. As a trusted source in the technology and IT sector, we provide a comprehensive resource for executives and professionals seeking to stay informed and ahead of the curve. With a focus on cutting-edge developments and trends, CIO News serves as your go-to destination for staying abreast of the rapidly evolving landscape of technology and IT. Founded in June 2020, CIO News has rapidly evolved with ambitious growth plans to expand globally, targeting markets in the Middle East & Africa, ASEAN, USA, and the UK.
CIO News is a proprietary of Mercadeo Multiventures Pvt Ltd.
