Salesforce Inc. on Thursday introduced observability features within its Agentforce Studio to give organizations enhanced visibility and control of their artificial intelligence (AI) agents.
The new suite of tools, integrated into the Agentforce 360 Platform, addresses a growing need among enterprises racing to implement AI at scale. According to Salesforce’s 2025 CIO Trends Report, AI implementation has surged 282%, pushing companies toward what the tech giant calls becoming an “Agentic Enterprise.”
The announcement comes as organizations grapple with the need for transparency. As agents handle critical customer interactions and revenue-impacting decisions, businesses require clear metrics on performance, learning capabilities, and improvement trajectories, Madhav Thattai, senior vice president and chief operating officer of Agentforce.
“Observability has become the new mandate,” Thattai said in an interview, emphasizing that Salesforce’s technology transforms uncertainty into measurable trust and return on investment. He added that about 12,000 of Salesforce’s customers have deployed Agentforce in some capacity.
The observability platform centers on three core capabilities built into Agentforce Studio and powered by Salesforce’s Session Tracing Data Model, which captures granular, session-level data within Data 360.
Agent Analytics provides performance tracking across all agents deployed. The feature translates agent activity into actionable data, surfacing key performance indicator trends over time and highlighting ineffective topics or workflows that require optimization.
Agent Optimization offers end-to-end transparency into every agent interaction. This capability lets teams trace session flows, understand decision-making processes, and diagnose performance gaps. The system can cluster similar requests to identify patterns and friction points while scoring responses based on intent, topic relevance, and quality metrics. Teams can pinpoint configuration issues and determine where tuning, retraining, or additional guardrails are necessary.
Health Monitoring is designed to ensure continuous operational reliability by detecting issues before they escalate. Using pre-calculated health metrics within Data 360, the system monitors silent failures that might otherwise go unnoticed. Supervisors receive proactive alerts on errors, latency issues, and escalations, enabling rapid problem resolution and minimizing downtime.
The platform’s unified approach aims to foster better collaboration between human teams and AI agents, ensuring AI adoption remains both reliable and trustworthy.
Tax preparation firm 1-800Accountant has deployed Salesforce’s Agentforce AI agents as a 24/7 digital workforce to handle customer inquiries and administrative tasks alongside human staff.
The AI agents tap into integrated data from Salesforce Data Cloud, Google documents, internal audit logs, customer support history, and external sources like the IRS website to provide instant, personalized responses to tax questions. Acting as the first point of contact, the agents autonomously schedule appointments and answer complex inquiries about deduction eligibility and tax return status without human intervention while also powering more than a dozen other workflows including lead recommendations, agenda automation, and client issue identification, Ryan Teeples, chief strategy officer at 1-800Accountant, said in an email.
