Agentic AI is rapidly emerging as the next phase of enterprise automation, moving beyond static workflows and copilots toward systems capable of autonomous reasoning, decision-making and action. Enterprises are increasingly experimenting with AI agents to augment customer service, IT operations and business processes, yet many struggle to operationalize these systems at scale. The challenge is no longer simply building agents, but ensuring they can be tested, governed, orchestrated and continuously improved within complex enterprise environments. Against this backdrop, Salesforce is addressing these challenges through its Agentforce platform, aiming to provide a comprehensive foundation for agentic AI across the enterprise.
Salesforce underscored this ambition by introducing a series of new capabilities designed to operationalize agentic AI at its recent TrailblazerDX (TDX) 2026 event, a more developer-focused counterpart to Dreamforce. These announcements reflect a broader market transition from experimentation to execution, where enterprises are seeking structured approaches to managing the full lifecycle of AI agents. Marking the event’s 10-year anniversary, Salesforce used TDX 2026 to reinforce its view that AI represents the most significant technological shift of the current era and to demonstrate how Agentforce is evolving into a comprehensive agent platform. The company introduced several key innovations, including Salesforce Headless 360, enhancements to Agent Fabric and deeper integration with Slack as a primary interface for agent interaction.
Headless 360 reflects Salesforce’s effort to decouple front-end experiences from back-end services, enabling developers to build and deploy agent-driven applications across multiple channels. The Agentforce platform is also being extended with new capabilities across the agent lifecycle. These include AgentScript, which introduces a hybrid reasoning model combining deterministic and probabilistic approaches, allowing enterprises to define structured workflows while still leveraging the flexibility of large language models.
Salesforce also highlighted new testing and evaluation tools, deployment options and observability features, including session tracing and telemetry. Early-stage capabilities including A/B testing APIs and multi-agent orchestration (currently in beta) will enable enterprises to manage fleets of agents rather than individual agents. The introduction of the Agentforce Experience Layer (AXL) enables agent deployment across multiple surfaces and user experiences, including Slack, Microsoft Teams and other conversational interfaces.
Salesforce is well established in agentic AI and was rated Exemplary for its AI Agents in our recent AI Buyers Guides. But the AI landscape is shifting from building individual agents to managing the full agent development lifecycle (ADLC). For enterprise buyers, including CIOs, application leaders, AI/ML specialists and customer experience executives, this evolution addresses a growing realization that isolated agent deployments are insufficient for long-term value.
Most enterprises today remain in the early stages of agent adoption. ISG’s Market Lens research of 1,200 top AI use cases shows that only 31% are in production. These
AgentScript’s graph-based reasoning model, with pre- and post-execution hooks, reflects an important advancement in making agent behavior more transparent and controllable. This is particularly relevant for regulated industries and large enterprises, where explainability and auditability are critical. Similarly, the introduction of observability tools and telemetry enables organizations to monitor agent performance, understand failure modes and iteratively improve outcomes.
The emergence of multi-agent systems is another key development. While still early, Salesforce’s investment in orchestration capabilities aligns with broader market trends. Enterprises are beginning to recognize that value will come not from a single agent, but from coordinated networks of agents working together across business processes. However, many organizations lack a clear understanding of what is required to implement such systems effectively.
Salesforce’s TDX 2026 announcements highlight the growing importance of managing the full agent development lifecycle. While Salesforce was rated Exemplary for its AI Agents in our Buyers Guides, it was not rated as highly in AI Governance and Operations. Enterprises need a structured approach to agent adoption, starting with low-risk, high-value use cases while building the foundations for scalability. This includes investing in testing, evaluation, governance and observability capabilities from the outset. These new capabilities should improve Salesforce’s rating in our assessment.
Organizations should also prepare for a shift toward multi-agent systems, ensuring that data readiness and integration strategies are in place. Evaluating platforms like Agentforce should involve not only their ability to build agents, but also their capacity to orchestrate and manage them at scale.
Looking ahead, Salesforce is likely to continue expanding its ecosystem, improving interoperability and enhancing its control plane capabilities. The bottom line for enterprises is clear: success with agentic AI will depend less on individual agents and more on the ability to manage the entire lifecycle.
Regards,
David Menninger