AI in Enterprise has moved beyond experimentation. That was one of the clearest takeaways from Dreamforce 2026, which concluded last week in San Francisco.
The keynote presented a broader vision for how enterprises can use AI across data, applications, agents and interfaces. Rather than treating AI as a standalone technology, Salesforce positioned it as a layer that can work with the systems, information and business processes an organisation already relies on.
For enterprise technology leaders, that shift matters.
The conversation is no longer only about which AI model to use. Instead, it is increasingly about how AI understands the business, how agents perform work, how enterprise applications provide context, and how organisations maintain trust and control as AI adoption expands.
The Dreamforce 2026 keynote provides a useful view of where that transition is heading.
What Dreamforce 2026 Revealed About AI in Enterprise
A central theme of the keynote was the gap between general AI intelligence and enterprise context.
AI models can understand a vast amount of world knowledge. However, they do not automatically know an organisation’s customers, products, transactions, service history, permissions, workflows, or business rules.
That distinction creates an important requirement for enterprise AI.
AI needs to be grounded in the information and systems that define how a business actually operates.
The keynote described this as bringing together probabilistic intelligence with deterministic enterprise intelligence. In simple terms, AI can provide reasoning and flexibility, while enterprise data and applications provide the trusted context, rules, and business knowledge required for reliable execution.
This shift changes how organisations should evaluate AI initiatives.
A model may be highly capable. Yet, without relevant business context, that capability does not automatically translate into enterprise value.
Enterprise AI Is Moving From Answers to Actions
The Dreamforce keynote also showed how enterprise AI is moving beyond generating responses.
Salesforce demonstrated agents working across sales, service, marketing, IT, HR, and supply chain processes.
The distinction is important.
A traditional AI assistant might help an employee find information or draft content. An enterprise agent can go further by interpreting business context, following defined processes, and taking actions within an organisation’s systems.
For example, the keynote demonstrated Marshall, an operations agent used for supplier onboarding. The agent learned a process in a controlled environment, understood the relevant business rules, and created a library of trusted actions before executing the process against SAP.
The demonstration illustrated a broader concept: AI reasoning needs to connect with reliable business execution.
That means the value of an enterprise agent depends not only on what it can understand, but also on what it can safely and consistently do.
Data Is the Foundation of Enterprise AI
One of the strongest messages from Dreamforce 2026 was that enterprise AI starts with data.
The keynote described enterprise context as the information that makes one business different from another. That includes customers, products, employees, transactions, conversations, and workflows.
However, this information often exists across multiple systems.
Data may sit inside applications, warehouses, data lakes, and other storage environments. Therefore, enterprises need to make that information accessible, trusted, and meaningful before agents can use it effectively.
The keynote positioned Informatica around preparing enterprise data for AI, and Data 360 around creating trusted enterprise context.
This distinction is useful.
Data is the raw foundation. Enterprise context is what makes that data meaningful to an AI system.
For example, knowing that a customer opened a support case is one thing. Understanding why the customer contacted the organisation, what they are trying to accomplish, and what action an agent can take next requires broader context.
That context is what allows AI to move from information retrieval towards useful business decisions.
For organisations building this foundation, Data Cloud can form part of the wider Salesforce data architecture.
Applications Are Becoming Part of the AI Layer
Data alone does not explain how an organisation operates.
Enterprise applications also contain business logic, relationships, permissions, processes and semantics built over years of implementation.
Dreamforce 2026 highlighted this application intelligence as another important layer of enterprise AI.
The keynote described Salesforce applications as increasingly headless, allowing their capabilities and business logic to be accessed through different interfaces.
This creates an important relationship between AI and enterprise software.
AI does not necessarily need to replace the application. Instead, it can use the application’s underlying capabilities while providing a different way for people and agents to interact with them.
That creates an architecture where:
Data provides context.
Applications provide business capabilities.
Agents reason and act.
Interfaces provide access.
The keynote brought these four layers together as the foundation for a new form of enterprise intelligence.
AI Agents Are Becoming a Digital Workforce
Agentforce was another major focus of the keynote.
Salesforce demonstrated agents designed for different business functions, including sales qualification, customer service, IT and HR services, marketing, commerce and supply chain operations.
The demonstrations showed a move towards a broader digital workforce.
Instead of one general-purpose agent attempting to handle every task, organisations can use specialised agents for specific processes.
For example:
- Sales agents can support lead qualification and pipeline generation.
- Service agents can handle customer conversations.
- Marketing agents can create and iterate campaigns.
- Operations agents can support repetitive processes.
- IT and HR agents can assist employees.
- Commerce agents can support shopping experiences.
The underlying idea is that agents become embedded within business functions rather than remaining separate AI experiments.
Salesforce stated during the keynote that more than 30,000 customers were using Agentforce, alongside billions of agent interactions. These figures were presented by Salesforce during the event and should therefore be understood as company-reported figures.
For organisations evaluating Agentforce, the bigger question is therefore not simply whether an agent can be deployed.
It is where an agent can create measurable business value.
The Interface Is Changing Too
One of the more distinctive themes at Dreamforce 2026 was the transformation of the enterprise software interface.
Traditionally, employees interact with enterprise software by opening an application, navigating menus, finding information and then completing a workflow.
The keynote presented a different model.
AI interfaces can bring enterprise information and actions directly to the user.
The demonstration of AI Force showed how an interface could bring together pipeline information, service data and marketing information while allowing the user to take action across Salesforce and Slack.
This represents a significant change in how enterprise software can be accessed.
The applications and data remain underneath the experience. However, the interface becomes dynamic, intelligent, and adaptable to the user’s requirements.
In practical terms, employees may increasingly interact with enterprise systems by describing the outcome they want rather than manually navigating to the feature they need.
That could mean asking an AI interface to identify an at-risk opportunity, analyse the account, and communicate the next step to a team.
The technology underneath still matters.
However, the interaction model is changing.
Governance Becomes Essential as AI Scales
As organisations deploy more agents, governance becomes a much larger enterprise requirement.
One agent operating within a controlled use case is different from an environment containing hundreds of agents from multiple providers.
Dreamforce 2026 addressed this through new experiences focused on agent management and security.
Agent Fabric was presented as a management layer for discovering and managing agents across providers. The keynote showed agents from Salesforce, Microsoft, AWS, and Google being surfaced within the environment.
Salesforce Guardian was presented as a security capability for agents.
This points to a broader enterprise requirement.
Technology leaders need to know:
- Which agents exist
- Who owns them
- What data they use
- Which systems they access
- What skills they have
- What actions they can perform
- Where their grounding data comes from
- How their activity is monitored
- How their permissions are managed
Agent governance therefore becomes part of the operating model for enterprise AI rather than simply a security feature.
The keynote also demonstrated agent lineage, including visibility into the data used to ground an agent. That connection between enterprise context and agent behaviour is particularly important as organisations move towards larger digital workforces.
AI in Enterprise Needs Trust, Not Just Intelligence
Another important theme from Dreamforce 2026 was trust.
The keynote repeatedly connected enterprise AI with security, governance and controlled access.
Salesforce also emphasised zero data retention in its discussion of AI Force, stating that customer data is not used to train other models when customers use its products. This is a Salesforce-stated product position from the keynote rather than an independent assessment.
The broader lesson applies beyond one platform.
Enterprise AI needs clear answers to questions around:
Where does the data go?
Who can access it?
Which model processes it?
What can an agent do with it?
How is its activity monitored?
What happens when something goes wrong?
Trust therefore needs to be designed into the architecture.
It cannot simply be added after an agent reaches production.
From AI Pilots to Business Processes
Perhaps the most important lesson from Dreamforce 2026 is the movement from AI experimentation towards business execution.
The keynote featured examples from sales, customer service, supply chain, recruiting, and other business functions.
In each case, the emphasis was not simply on generating an impressive AI response.
The focus was on what the AI could accomplish.
For example, the supplier onboarding demonstration showed how an agent could learn a repeatable process and then execute it against a back-end system.
Similarly, the recruiting demonstration showed an agent engaging candidates at scale while business leaders monitored its impact on the hiring funnel.
This changes how organisations should measure enterprise AI.
Instead of asking only:
“How intelligent is the model?”
business leaders increasingly need to ask:
“What business process can this system improve?”
That shift brings AI closer to measurable outcomes such as productivity, customer experience, operational efficiency and revenue generation.
What Dreamforce 2026 Means for Enterprise Technology Leaders
The keynote points to several practical considerations for organisations developing their AI strategy.
1. Start With Enterprise Context
AI becomes more useful when it understands the organisation’s customers, data, processes and business rules.
2. Treat Data as an AI Foundation
Data quality, accessibility, governance and semantics directly affect the usefulness of enterprise AI.
3. Connect Agents to Real Business Processes
The strongest use cases are not isolated conversations. They connect AI reasoning with defined business workflows and actions.
4. Plan for Multiple Agents
Organisations should consider how agents will be discovered, managed, secured, and governed as adoption grows.
5. Rethink the User Interface
AI can provide a new interaction layer for enterprise applications, reducing the need for users to navigate complex software manually.
6. Measure Business Outcomes
AI adoption should ultimately connect to business performance rather than technology adoption alone.
Where NJC Labs Fits Into the Enterprise AI Shift
For NJC Labs, the message from Dreamforce 2026 reinforces a practical principle: enterprise AI needs to work with the systems and processes that already run the business.
Agents need trusted data to understand context. They need access to business applications to perform meaningful actions. They need defined processes to execute consistently. And they need governance to ensure those actions remain within enterprise boundaries.
That makes implementation just as important as the AI experience.
NJC Labs works across Salesforce, Agentforce, Data Cloud and MuleSoft to help organisations put these technologies into practical use.
The objective is not simply to introduce another AI tool.
Instead, the focus is on creating connected enterprise environments where AI can work with business data, applications and processes to deliver measurable outcomes.
That is the practical path towards intelligent transformation.
The Bigger Takeaway From Dreamforce 2026
Dreamforce 2026 showed that enterprise AI is entering a more connected phase.
The focus has moved beyond the AI model itself.
The model is one part of the architecture.
Around it, enterprises need trusted data, business context, application intelligence, specialised agents, intelligent interfaces and governance.
That creates a broader enterprise AI model:
Data → Context → Applications → Agents → Actions → Outcomes
Each layer has a distinct role. Trusted data gives AI the context it needs to operate effectively. Application intelligence helps agents understand business processes and requirements. Agents turn that intelligence into action, while governance provides the control needed to scale AI responsibly. Ultimately, business processes connect these capabilities to measurable outcomes.
The significance of Dreamforce 2026 is therefore not simply the introduction of new Salesforce products.
It is the growing convergence of AI, enterprise data, business applications and digital workforces.
For organisations already experimenting with AI, the next step is increasingly about connecting those experiments to the systems and processes that create business value.
About Dreamforce 2026
Dreamforce 2026 concluded in San Francisco on September 17, 2026, following the September 15-17 event.
The event’s keynote focused heavily on enterprise AI, Agentforce, enterprise context, AI Force, intelligent interfaces, agent management, and the emerging agentic enterprise.
For official event information and Salesforce’s event resources, visit Dreamforce 2026.