Salesforce has introduced new tools to streamline the lifecycle management of AI agents developed with Agentforce. The new tools streamline testing, deployment, and monitoring processes while offering transparency in usage and performance.
Designed for enterprises striving to adopt an “agent-first” approach, the new toolchain enables teams to safely experiment and scale their AI capabilities without risking disruptions to live systems.
The new Agentforce Testing Center leverages the Salesforce Platform and Data Cloud integration to conduct large-scale, automated tests. Teams can generate hundreds of synthetic interactions to evaluate the agents’ ability to handle varied customer queries effectively, refining instructions to improve their accuracy and reliability.
Secure development with Sandboxes
Salesforce Sandboxes now extend their capabilities to support Data Cloud and Agentforce, providing isolated environments for safe testing. These sandboxes mirror production data and configurations, enabling development teams to prototype and validate AI agents rigorously. This includes user acceptance testing to confirm agents meet business requirements before deployment.
The seamless migration of these configurations to production is supported through familiar tools like Change Sets and DevOps Center, enhancing efficiency and minimising risk.
“Agentforce is helping businesses create a limitless workforce. To deliver this value fast, CIOs need new tools for testing and monitoring agentic systems. Salesforce introduced the concept of Application Lifecycle Management back in 2006 with Force.com. This new category of Agentic Lifecycle Management requires unique tools, and Salesforce is meeting the moment again with Agentforce Testing Center, which will help companies roll out trusted AI agents with no-code tools for testing, deploying, and monitoring in a secure, repeatable way,” said Adam Evans, EVP and GM for Salesforce AI Platform.
New monitoring solutions, including Agentforce Analytics and Utterance Analysis, offer detailed insights into AI agent performance. These tools, built on Data Cloud, allow continuous improvement of agents through user feedback and prompt refinement.
Additionally, usage tracking is enhanced via Digital Wallet, which now provides granular visibility into feature consumption. Automated alerts can notify teams of unexpected usage spikes, helping enterprises manage resources more effectively.