As the AI agent workforce grows, Salesforce offers more control and more predictability in AI operations.
The rapid expansion of AI agent populations within enterprise environments has created a pressing need for robust management and governance solutions. Recognizing this challenge, Salesforce is significantly enhancing its Agent Fabric platform, a comprehensive suite designed to address "agent sprawl" and bring much-needed predictability to complex AI operations. The latest updates aim to provide organizations with finer-grained control over their AI agents, whether they are native to the Salesforce ecosystem or external solutions, thereby mitigating risks associated with redundancy, silos, and operational inefficiencies.
The genesis of Salesforce’s proactive approach to AI agent management dates back to last year with the introduction of Agent Fabric, initially integrated within its MuleSoft AnyPoint Platform. This foundational suite was conceived as a centralized registry for enterprises to catalog, monitor, interconnect, and govern their burgeoning AI agents. The platform’s evolution has been marked by incremental yet crucial additions. In September 2025, Agent Fabric was launched with the core functionalities for managing AI agents. This was followed by significant enhancements in January of the current year, which included the introduction of a deterministic scripting tool, allowing for more predictable agent behavior, and the capability to scan for and automatically register new AI agents entering the enterprise landscape.
However, the persistent challenge of scaling AI safely and responsibly has necessitated further advancements. Salesforce’s latest suite of features is designed to provide enterprises with enhanced control mechanisms, particularly in orchestrating the complex interactions between multiple AI agents and in managing the associated costs of large language model (LLM) usage.
Deterministic Controls for Predictable AI Workflows
A key innovation is the expansion of deterministic controls through "Agent Script for Agent Broker." Agent Broker, an intelligent routing service within Agent Fabric, is designed to dynamically connect AI agents across various domains, matching user tasks with the most suitable agent. The new Agent Script feature empowers developers to codify specific workflows within multi-agent systems. This allows for the predefinition of rules and sequences of operations, ensuring consistent and reliable outputs, a critical factor for enterprise adoption.
Traditionally, probabilistic agents, while offering flexibility and advanced reasoning capabilities, can introduce an element of unpredictability into operations. By integrating Agent Script for Agent Broker, Salesforce is enabling enterprises to steer decision-making processes according to predetermined logic. This approach not only enhances predictability but also offers potential cost efficiencies. Executing predefined rules often requires fewer computing resources compared to relying solely on the extensive processing power of large language models for every decision point.
Robert Kramer, Managing Partner at KramerERP, lauded these advancements, stating, "Pure autonomous agents don’t necessarily work in production as enterprises need to ensure predictable outcomes. The deterministic controls should facilitate a secure handoff of control and rules while still allowing the model to engage in reasoning when it’s appropriate. It’s a balance between control and flexibility, which is the norm for most real deployments." This sentiment highlights the practical need for a hybrid approach that leverages the strengths of both deterministic and probabilistic AI.
Rebecca Wettemann, Principal Analyst at Valoir, echoed this view, emphasizing the strategic advantage of offering both deterministic and probabilistic options within a single platform. "Providing both deterministic and probabilistic options within Agent Fabric enables developers and agent builders to take the lower-cost route to more accurate and predictable results from agentic systems," she commented.
Despite the anticipated benefits, enterprises will need to exercise patience as this advanced orchestration feature is currently in beta testing. General availability is slated for June 2026, indicating a deliberate and thorough validation process by Salesforce.
Centralized LLM Governance for Cost Management
Beyond workflow orchestration, Salesforce is introducing a significant new capability: LLM Governance within its AI Gateway. AI Gateway serves as the central control layer for Agent Fabric, providing comprehensive visibility into token usage, associated costs, and data flows for third-party LLMs. The newly available LLM Governance feature is designed to equip enterprises with the tools to effectively manage and maintain their AI operations within budget constraints.
This development arrives at a critical juncture as Chief Information Officers (CIOs) grapple with the escalating costs of AI initiatives and the imperative to centralize control over disparate AI systems. The unchecked proliferation of AI models, each potentially managed with unique API contracts and token budgets by different teams, can lead to significant cost overruns and security vulnerabilities.
Scott Bickley, Advisory Fellow at Info-Tech Research Group, cautioned about the implications of a lack of centralized governance. "Without centralized governance like this, different teams around a company may choose different models, negotiate their own API contracts, and manage token budgets locally. This results in sprawling costs, inconsistent security postures, and no enterprise-wide policy enforcement," he explained. Bickley further elaborated on the benefits of AI Gateway acting as a central "choke point" for all LLM traffic. "By positioning AI Gateway as the choke point through which all LLM traffic flows, enterprises gain visibility into AI usage patterns, the models in use, purpose of the usage, and cost data," he added. This centralized visibility is crucial for accountability and strategic resource allocation in AI deployment.
Streamlined Integration with MCP Additions
To further simplify the integration of AI agents with existing enterprise infrastructure, Salesforce is enhancing its Model Control Protocol (MCP) features. The introduction of MCP Bridge aims to facilitate smoother access to legacy APIs, while Informatica-hosted MCPs are designed to simplify agent interactions with enterprise data and existing APIs. These additions are poised to significantly reduce development time and streamline the creation of complex, cross-environment, multi-agent systems.
Scott Bickley highlighted the practical impact of MCP Bridge, particularly for organizations burdened with a vast number of legacy APIs, including REST, SOAP, and GraphQL, developed long before the advent of MCP. "Agents speaking MCP cannot call those APIs natively so they require wrappers around the API endpoint; this would be a massive engineering lift. MCP Bridge allows these APIs to be exposed as MCP-compatible tools without modifying the underlying code," he stated. This capability effectively bridges the gap between modern AI agents and decades-old enterprise systems, preventing costly and time-consuming refactoring efforts.
Rebecca Wettemann emphasized the value of Informatica-hosted MCPs in reducing development overhead by embedding data quality and governance capabilities directly into agent workflows. This is particularly critical for enterprises operating in regulated industries or those with heightened risk concerns, where data integrity and compliance are paramount.
However, Bickley also introduced a note of caution regarding the practical implementation of these integration features. "APIs can behave oddly and have their own nuanced behavior," he warned. "Enterprises should test how MCP Bridge handles edge cases." He also tempered expectations regarding Informatica-hosted MCPs, noting that while data quality and governance capabilities are valuable, they are not instantaneous. "Even if the Informatica data quality and governance capabilities are cleanly integrated in the Agent Fabric registry, these are not instantaneous operations. Checking data fields for accuracy, deduplication, and cross-system matching take time and carry latency measured in milliseconds or even multiple seconds, and that is pre-integration," he pointed out. This suggests that while integration is simplified, performance considerations for data processing remain crucial.
A Strategic Pivot for MuleSoft
Industry analysts perceive these extensive updates to Agent Fabric as indicative of a broader strategic repositioning for MuleSoft, Salesforce’s acquisition from 2018 for $5.7 billion. The company appears to be transitioning MuleSoft from its traditional role as an API integration platform to becoming a foundational infrastructure layer for enterprise AI agents.
By embedding orchestration, governance, and robust connectivity solutions within Agent Fabric, Salesforce is positioning MuleSoft as the definitive system of record for how AI agents are discovered, routed, and governed across an organization. This evolution aims to deepen MuleSoft’s significance beyond mere API management, establishing it as a core component of an enterprise’s AI infrastructure.
This strategic shift, however, may not be universally welcomed by all CIOs. The increasing integration of critical AI control plane functionalities within Agent Fabric could lead to significant switching costs. The more agents an organization registers, and the more complex its orchestration rules and governance policies become, the more challenging it will be to migrate to an alternative solution.
As with any critical infrastructure dependency, CIOs are advised to carefully consider the long-term implications. Key questions to address include the exit strategy, the portability of Agent Fabric components versus vendor lock-in, the pricing model’s long-term sustainability, and the depth of integration with non-Salesforce agents and data sources. This due diligence is essential to ensure that the adoption of advanced AI management platforms aligns with an organization’s strategic agility and risk management objectives. The evolution of Agent Fabric signals a significant step towards managing the complex and rapidly growing landscape of enterprise AI agents, offering enhanced control and predictability as organizations continue to embrace artificial intelligence at scale.
