From Language to Action: How Salesforce is Redefining AI in CRM

The tech world is abuzz with short forms like LLMs, SLMs, and LAMs. But what exactly do these terms mean, and how are they different from each other? Let’s break it down for you.?

?

Real-World Scenario: The Sales Agent’s Day?

Imagine a sales agent who has been following up on a big opportunity. As his day begins, he wants a quick summary of recent conversations and quote updates to ensure no detail is missed that could make or break the deal. He pulls out his mobile, navigates to the chat assistant in his CRM app, and says, “Give me a summary of the Ohana Enterprise opportunity.” The assistant responds promptly with: “Here is a summary of the Ohana Enterprise opportunity: ........”?

In the background, the assistant is powered by either an LLM (Large Language Model) or an SLM (Small Language Model). LLMs are deep learning models with billions of parameters, trained on vast amounts of text data. These models can understand and generate human-like language, excelling at tasks like text summarization, translation, and question-answering. SLMs, on the other hand, are lightweight versions of LLMs, designed for more specific, targeted tasks, often with fewer computational requirements.?

Taking Action with Large Action Models (LAMs)?

Now, our sales agent realizes that Ohana Enterprise has shown interest in a vendor management tool, which happens to be a perfect fit for what they offer. He wants to send a follow-up email and arrange an internal meeting to align the team’s strategy for this opportunity.?

With just a single sentence in plain English, he can accomplish this. LAMs (Large Action Models) come into play here. These models enable the assistant, now functioning as a proactive agent, to perform tasks autonomously without the need for step-by-step human guidance. The assistant drafts an email to the client, summarizing action items from the last meeting and proposing a vendor management tool. It also schedules an internal meeting with the relevant tech experts—all without further input.?

While LLMs analyze user input and generate contextually relevant responses, LAMs take this interaction a step further by managing entire workflows. They perform complex actions and automate processes, allowing users to focus on higher-level tasks rather than navigating multiple steps manually.?

?

?

Salesforce’s Evolution in AI-Driven CRM?

Since 2014, Salesforce has led the AI revolution in CRM with Einstein—a comprehensive suite of AI-driven solutions designed to supercharge business processes. The launch of the Einstein 1 Platform, which includes tools like Prompt Builder, Model Builder, and Skill Builder, has solidified Salesforce’s position as the go-to platform for integrating AI across sales and service functions.?

Recently, Salesforce unveiled its proprietary family of "Large Action Models" called xLAM, marking a significant leap forward in AI-driven automation. Another in-house model, xGen-Sales, has been developed to empower autonomous sales tasks on Agentforce, the platform that powers Salesforce’s AI agents. The announcement follows Salesforce’s strategic acquisition of Airkit.ai , a company known for its eCommerce AI expertise.?

Salesforce now boasts its own Sales Agent, Sales Development Rep (SDR) Agent, Sales Coach Agent, and Service Agent, all equipped with advanced function-calling and multi-modal capabilities. These AI agents, backed by Salesforce’s in-house models, help drive smarter, more efficient customer interactions. Notably, xLAM 8x22b has claimed the top spot on the Berkeley Leaderboards for function calling, outperforming even GPT-4.?

In line with its commitment to data security, privacy, and compliance, Salesforce made another major move on September 5th, 2024, with the acquisition of Own Company, a leading provider of data protection and management solutions. The Own Data Platform offers robust capabilities for data archiving, seeding, security, and analytics—ensuring that Salesforce customers can maintain the highest standards of data compliance and protection across their mission-critical SaaS environments.?

?

As Salesforce continues to push the boundaries of AI-driven CRM with innovations like the Einstein 1 Platform and xLAM models, it’s clear that the future of business automation is not just about managing customer relationships—it's about revolutionizing how businesses engage, sell, and service through intelligent, autonomous systems that prioritize efficiency, security, and personalized experiences.?

?Article Credit & Contribution : Also contributed by Shravani Nevagi , Lead AI Engineer Appstrail Technology