Singapore – Salesforce has unveiled a new architecture designed to give artificial intelligence agents a shared, trusted understanding of a business and its customers, alongside a new AI Control Plane to manage AI activity across the enterprise.
The company announced the Enterprise AI Harness on 11 September 2026 in Singapore, positioning it as the foundation businesses need as AI agents take on increasingly complex roles across every function of an organisation.
According to Salesforce, the rise of what it calls the “Agentic Enterprise” is changing how work gets done, with agents now expected to understand situations, decide on next steps, act across systems, and collaborate with both people and other agents.
That shift, the company says, creates a fresh challenge for enterprises: equipping agents with what they need to operate reliably, securely, and at scale.
The Enterprise AI Harness is Salesforce’s answer to that challenge. Salesforce describes it as a trusted foundation that brings together six capabilities spanning context, agency, action, governance, security, and models.
These capabilities are delivered through what the company calls a common, composable architecture, built on the customer relationships, processes, and controls already running within a business.
Salesforce says a new AI Control Plane accompanies the six capabilities, giving businesses a single place to see, manage, and control agents and AI as they spread across the enterprise.
Customers, the company adds, can use the six capabilities together as one system or adopt only what they need, whether through Salesforce technology, their existing technology, or a combination of both, including third-party models, agents, and systems.
To illustrate why the architecture matters, Salesforce points to a seemingly simple customer question: “Can we fulfil this order today?”
No single system holds the complete answer, the company explains. CRM knows the customer and the relationship, ERP knows inventory and fulfilment, and contracts contain commitments and entitlements.
Analytics provides business definitions, policies determine what can be promised, and previous interactions provide critical memory, Salesforce notes, while business processes determine how the work should actually get done.
The challenge, according to the company, is not simply connecting these systems. An agent needs to understand all of that information in the context of a specific customer at a specific moment, then determine the right action, take it securely, and operate within the business’s policies and controls.
Salesforce argues that while AI reasoning can be open-ended, enterprise execution often cannot be, and the Enterprise AI Harness is designed to connect that reasoning to the business rules required for predictable outcomes.
By making these capabilities reusable across the enterprise, the company says context can be shared across agents and models, actions and workflows can be securely invoked wherever needed, and governance and security can be applied consistently as AI moves across the business.
As people and agents work together, Salesforce says every interaction and action can also generate new signals, outcomes, and memory that enrich the context available for what comes next, creating what it describes as a compounding intelligence loop over time.
Models will keep improving and becoming more widely available, the company acknowledges, but it maintains that an enterprise’s proprietary context, including its customers, knowledge, semantics, memory, processes, relationships, and history, remains uniquely its own and becomes a durable source of differentiation as underlying intelligence evolves.
Salesforce says the Enterprise AI Harness draws on core technologies from across Data 360, Informatica, MuleSoft and Agent Fabric, Tableau, Agentforce, Salesforce Guardian, and the Salesforce Platform, unified through a common, composable architecture.
For existing Salesforce customers, the company says the technologies they already use will provide the foundation for this expansion, with an upgrade path to unlock new capabilities as they become available.
Turning to the six capabilities themselves, Salesforce describes Trusted Context as bringing together customer context with data, metadata, semantics, knowledge, real-time signals, memory, and an understanding of how work gets done across the enterprise.
Grounded in governed enterprise data, the company says Trusted Context gives AI a shared understanding of the customer and the business around them, including, in the order example, the customer’s contract, entitlements, and available inventory.
Trusted Agency, meanwhile, is described as giving agents the reasoning, planning, state, memory, collaboration, and orchestration needed to pursue complex business outcomes, combining flexible AI reasoning with deterministic controls where certainty is required.
Trusted Action securely connects AI to applications, APIs, workflows, tools, and business processes, according to Salesforce, allowing agents to move from understanding what needs to happen to actually getting it done, with outcomes flowing back into Trusted Context.
Trusted Governance, the company says, helps enterprises govern the data, metadata, policies, and processes AI relies on, with lineage, quality, guardrails, and controls to ensure agents follow the correct policies.
Trusted Security applies identity, permissions, privacy, data protection, and runtime security to what AI can access and what agents are permitted to do, Salesforce explains, ensuring agents can only see authorised customer information and take authorised actions.
Rounding out the six, Trusted Models gives enterprises the flexibility to securely connect the right model for each job, with intelligent model routing based on accuracy, performance, cost, and business requirements, the company says.
“The Agentic Enterprise won’t be defined by which model a company chooses. Models will continue to change, and intelligence will increasingly be available everywhere. What will differentiate an enterprise is the trusted, proprietary context it brings to that intelligence — starting with the customer — and its ability to securely turn that context into action,” said Rohan Kumar, President, Chief Platform and Engineering Officer at Salesforce.
“That’s what we’re building with Salesforce’s Enterprise AI Harness: a system where every interaction and action can create new context and intelligence, making the enterprise’s context richer over time while operating with the security, governance, and control businesses need to trust AI at scale,” Kumar added.
On the newly introduced AI Control Plane, Salesforce says businesses need a consistent way to know what AI is operating across their organisation, how it is performing, what it is allowed to do, and how much it costs, as agents and AI continue to spread across the enterprise.
The AI Control Plane is intended to give businesses a common place to see, manage, and control agents and AI across the enterprise, enabling companies to discover and register agents and AI capabilities, establish identity and policy, manage lifecycle, evaluate performance, observe behaviour and outcomes, and control cost across both Salesforce and third-party AI.
This, according to the company, gives enterprises a consistent layer of visibility and control as AI expands across teams, applications, models, and systems, without requiring every agent or AI experience to be managed separately.
Salesforce says the Enterprise AI Harness has been designed to meet customers wherever and however they choose to build and use AI, with a new unified, modern experience intended to make it easier for different teams, from data and security leaders to builders, to work with the capabilities most relevant to them.
The architecture is also being built headlessly from the ground up, the company says, with capabilities accessible through technologies including MCP, APIs, Skills, and Plug-ins.
This, Salesforce explains, allows its capabilities to extend beyond traditional Salesforce applications and into the AI experiences where people already work, including Claude, Slack, Microsoft Teams, and Agentforce, in support of the company’s broader AIforce strategy to bring Salesforce to any AI, agent, app, or surface.
“AI is moving faster than any technology we’ve seen — which is exactly why composability matters so much to us. We don’t want to bet our future on one closed stack; we want the freedom to adapt as the landscape shifts. That’s what resonated with us about Salesforce’s new Enterprise AI Harness,” said Shawn Malhotra, CTO, Rocket Mortgage.
