Dreamforce 2026: from using AI to putting it to work in the business

Agents that collaborate, data that provides context, and a new layer of enterprise orchestration. These are the changes that really deserve your attention after Dreamforce 2026.

For the past few years, much of the conversation around artificial intelligence has centered on one question: what can AI do?

Dreamforce 2026 started shifting that conversation toward a far more important one: how do we get AI to actually work inside the business?

That shift might seem small, but it changes the way we think about technology quite a bit. The conversation is moving toward a business where agents, people, applications, and systems can work in a more coordinated way, with access to the context needed to execute tasks, make decisions, and complete processes end to end. 

At VASS, we see this shift as a meaningful evolution because it significantly expands the role Salesforce can play within an organization. It's no longer just about managing the customer relationship. It's establishing itself as a platform capable of connecting context, data, processes, and intelligence to enable new ways of working.

The user experience decouples from the platform

One of the most visible changes is how people can interact with Salesforce. The evolution shown at Dreamforce points to a much more flexible experience. A user can interact through Slack, a conversational interface, or other channels, while Salesforce keeps operating behind the scenes with the information, permissions, rules, flows, and business logic needed to execute the action.

This opens up an interesting possibility for organizations: bringing technology capabilities to where people already work, reducing friction and making interaction with complex systems feel far more natural. A salesperson, for example, could request a pipeline analysis, identify stalled opportunities, or prepare context for an account through a conversation, without having to manually navigate different screens.

The value isn't just in changing the interface. It's in simplifying how people access the organization's knowledge and capabilities.

From individual agents to an orchestration logic

Early enterprise AI use cases mostly centered on assistants or agents capable of answering questions and executing specific tasks. Dreamforce 2026 showed a shift toward architectures where different agents can specialize in different functions and work in a coordinated way. One agent might retrieve information, another execute a process, another update data, and another interact with an external platform, with a layer above coordinating these capabilities to solve a broader need.

For the user, the experience can stay simple. For the organization, however, this means designing an architecture capable of connecting different sources of information, tools, and decision levels. Agents were shown for sales, marketing, service, procurement, and support, alongside operational use cases capable of interacting with systems like SAP, extending the conversation from the CRM into much broader enterprise processes.

Context becomes a critical capability

As an agent can do more, it also needs to better understand the environment it's operating in. An agent involved in a sales decision needs to understand who the customer is, their history, open opportunities, business rules, and previous interactions and that same principle applies to an operational agent that needs inventory data or transactional systems, or a service agent that needs to know past cases and the policies that determine what actions it can take.

That's why Data 360 takes on even greater importance within this new architecture. Context can come from Salesforce, but also from documents, backend systems, middleware, data lakes, warehouses, and other sources of information. The ability to connect that data, keep it governed, and make it accessible at the right moment becomes one of the foundations for AI to generate real value.

An example built around Formula 1 helps visualize this. Integrating information on browsing, purchases, event tickets, preferences, and engagement level makes it possible to build a much more complete view of the customer and use that context to enrich the relationship with the brand. AI's potential increases considerably once it stops working with isolated information and starts understanding the full context of each interaction.

More open architectures, faster development

Dreamforce also revealed a more open vision of the ecosystem: the coexistence of different channels, agents, applications, models, and platforms, connected through different integration and orchestration mechanisms. This matters because organizations rarely operate on a single technology stack. Their processes tend to be spread across CRM, ERP, data platforms, collaboration tools, in-house systems, and solutions from different vendors. The partnerships showcased with companies like Anthropic, NVIDIA, Google Cloud, and AWS reinforce this move toward more interoperable, multi-model environments.

From our perspective, this raises the importance of architecture design. The question is no longer just which technology to use, it's how to connect the different capabilities coherently.

At the same time, applying natural language to technical tasks is changing the speed of development. In demos built around MuleSoft, some integrations could be configured using conversational instructions, in significantly less time than traditional processes require. This can speed up certain stages of development, but it shifts the focus onto other capabilities that remain critical: architecture, data quality, security, testing, governance, and operations. Technology can help build faster, but it's still essential to design correctly what's being built and how it should work within the enterprise ecosystem.

More autonomy requires more control

Agent autonomy opens up significant opportunities, but it also increases the need to set clear boundaries. When an agent can query information, update systems, or execute processes, the organization needs to precisely define what it's authorized to do, what data it can use, and when a person needs to step in. That's why concepts like security, isolation, data masking, access control, and observability take center stage within this new architecture, and why the human-in-the-loop principle remains essential.

Not every process needs the same level of autonomy. Some tasks can run automatically, while others will require validation or human intervention depending on their impact and risk level. Getting that balance right will be a fundamental part of any enterprise AI strategy and it also explains why the conversation around Agentforce is moving from experimentation toward real consumption and adoption: after an initial phase of pilots, the challenge now is proving how these capabilities get embedded into processes and generate sustainable results.

For organizations, the starting point should be the use case

Given the speed at which new capabilities keep appearing, there's an obvious risk: starting the conversation from the technology. At VASS, we believe the starting point should still be the business. Before deciding which agent or tool to implement, it's worth understanding which process needs to improve, where the friction is, what outcome you want to generate, and what information the organization needs to get there. From there, you can design the right combination of data, agents, integrations, and platforms.

For a business leader, this can be translated into four concrete questions:

1. What process or decision do we want to improve?

2. What data and systems does the AI need to understand to do it correctly? 

3. What level of autonomy is right for that process?

4. How will we measure and improve the outcome once it's up and running?

Our POV: the opportunity lies in connecting the pieces

Dreamforce 2026 showed a significant shift in how enterprise AI is understood. Agents are gaining more capacity to execute, collaborate, and take part in real processes. Data is becoming even more important as a source of context. Architectures are becoming more open, and interacting with technology is starting to get simpler.

For us, the opportunity lies in connecting those capabilities around concrete business needs. The organizations that succeed won't necessarily be the ones that deploy the most agents or adopt the most tools, they'll be the ones that manage to properly integrate data, processes, technology, and governance to transform the way they work.

That's the space where we believe the next stage of enterprise transformation begins: designing AI capabilities that understand the organization's context, operate within its processes, and generate results that can be measured and evolve over time.

The question is no longer whether AI can do more. It's where it can generate real value for your business. 

If you're evaluating use cases with Salesforce, Agentforce, or Data 360, at VASS we can help you prioritize opportunities, lay the groundwork, and design a roadmap that connects technology with results.