Salesforce announced on Mar. 16 a new collaboration with NVIDIA to bring high-performance, cost-efficient artificial intelligence agents into the daily operations of large enterprises. The partnership aims to address the challenge of integrating AI agents with the systems, data sources, and workflows that organizations rely on for their business processes.
The initiative is significant for regulated industries that require secure, governed, and compliant AI solutions at scale. By combining Salesforce’s Agentforce platform, Slack, and NVIDIA’s Nemotron models from the NVIDIA Agent Toolkit, companies can now deploy AI agents in both regulated and on-premises environments while allowing employees to interact with these agents directly within their workflow.
At the center of this effort is Agentforce, Salesforce’s platform designed for building and managing enterprise AI agents. Agentforce connects these agents to trusted data through Data 360, applies business logic, and coordinates actions across Customer 360 applications as well as other connected systems. This ensures governance, compliance, and predictable performance at scale.
NVIDIA’s Nemotron 3 Nano model is integrated within Agentforce to provide optimized performance for enterprise needs. Its architecture allows it to process long customer histories and complex workflows efficiently by using a Mixture of Experts approach that reduces computational demand and costs. For industries with strict regulatory requirements such as financial services or healthcare, Nemotron models can be deployed on NVIDIA AI infrastructure within private cloud or on-premises environments so that sensitive data remains secure.
Slack serves as the interface connecting users with these AI agents. Slackbot acts as a coordination layer by receiving user requests in Slack channels, triggering Agentforce workflows, invoking Nemotron-powered processing when needed, and returning recommendations back into conversations. This setup enables use cases like compliance checks in finance or summarizing case histories in healthcare while maintaining security controls defined within Salesforce.
Salesforce and NVIDIA are also developing reference architectures that outline how each component—Slack for engagement, Slackbot for coordination, Agentforce for reasoning and execution, Data 360 for context management, and NVIDIA infrastructure for accelerated processing—fits together. These blueprints aim to remove ambiguity around deploying secure agent solutions at scale.
The companies say this collaboration will help move enterprise AI adoption from pilot projects to production by ensuring solutions are governed, scalable, cost-efficient, and embedded directly into real business workflows.
