Late last month Salesforce announced its first AI-powered agents, addressing business development and sales coaching use cases. Fast on that news is the release of new proprietary AI models, under the xGen moniker. These are essentially new libraries of large language models (LLMs) designed for generative text use cases supporting CRM processes. Think email and web content creation, document summaries, etc.
The latest xGen release is xGen sales, which the company says is “a proprietary model trained and designed to power autonomous sales tasks with Agentforce, and xLAM, a new family of Large Action Models designed to handle complex tasks and generate actionable outputs.” In short, Salesforce is full steam ahead in entering the “autonomous” phase of AI development, at least on paper. Its pre-packaged SDR and sales coaching AI agents are a safe, and solid choice for a first foray - but now the company is opening up the tool set to allow customers and parters to build more autonomous and action-oriented use cases for AI virtual agents.
It is a smart move. Putting both the onus on the customer and the partner to figure out the last mile of AI agent development allows for a wider proliferation of use cases, and also keeps Salesforce off the hook in some ways (but not all) when it comes to compliance and security.
These new AI agent tools will get a boost in the near future, now that Salesforce has announced its plans to acquire Tenyx, a developer of AI-powered voice agents. The deal is expected to close by the end of October, and the company says Tenyx’s technology will enable innovation inside its AI agent offerings (yet to be GA) around customer service and support use cases. The AI voice agent technology can quickly augment chatbot and other AI multi-channel tools by offering more intuitive and actionable virtual agent experiences, where these agents can make more decisions and be injected into more workflows to complete more tasks to drive productivity, time to resolution, etc.
As always, current Salesforce users should carefully inspect and evaluate these new LLMs and LAM offerings. Both for compliance and security, but also for alignment in terms of use case and expected business outcomes. The old adage “just because you can, does not mean you should” comes to mind this early in the development of these new AI agents. In short, proceed with optimistic caution.
