2026 Dreamforce was different.
For the past several years, I haven’t attended the event in person, choosing instead to watch the keynotes online while spending time at more focused co-located events that offer a closer pulse on how businesses are approaching sales and service.
I regret not attending in person this year. It felt like a special edition.
In recent years, Dreamforce has been coming across as a showcase of increasingly polished demos, with presentations tightly scripted. This year, the tone was more human and relatable. Presenters were more natural and conversational, while the content placed greater emphasis on practical, pragmatic steps for putting technology to work.
Let’s unpack the main announcements.
Salesforce realigned its offer around seven agents built for specific jobs. This is a pivot from last year’s broader positioning around becoming an agentic enterprise, a vision that remains difficult for most businesses to execute at this juncture:
Piper, inbound SDR
Hunter, outbound BDR
Casey and Fin, customer service
Carter, product FAQs for online retailers
Paige, IT and HR employee support
Marshall, supply chain and back-office workflows
This reinforces a theme I have highlighted in my research and coverage of Salesforce’s acquisitions of Qualified, Momentum, and Fin: purpose-built agents are proving to be the most practical approach today.
Salesforce introduced AIforce, a new experience layer that brings Salesforce capabilities into the applications and environments where people and agents work, including Claude, Slack, and more to come. It was the star of the show. AIforce builds on what Salesforce originally announced at TDX 2026 as Headless 360. The underlying Headless Toolkit provides APIs, MCPs, CLIs, Skills, and other capabilities that decouple Salesforce’s data and business logic from its UI, allowing both humans and agents to access Salesforce capabilities from wherever they work.
Data 360 is now positioned as an agentic CDP. I commend Salesforce for calling a cat a cat and using CDP as the product descriptor. Many companies have become reluctant to use the term, making data layers increasingly nebulous at a time when they have become more critical than ever. Salesforce introduced the Agent Context Engine (ACE), a delineated context layer for AI agents, whether Salesforce or third-party.
Salesforce also announced Koa, a CRM reasoning model designed to help agents reason through complex, multi-step workflows and select the right tools to get work done. Koa is trained on a proprietary synthetic dataset modeled on enterprise knowledge from nearly three decades of CRM deployments.
The Service keynote focused on breaking down application silos and enabling a hybrid workforce, illustrated through the Hybrid Workforce Command Center. In its pre-event product adoption update, Salesforce highlighted the strong adoption of service AI, with service use cases driving five times more AWUs than other use cases. Interestingly, Fin and Casey, while functionally similar, coexist with different roles, with Fin focused on speed and Casey on complexity. Another key announcement was Agent Optimizer, an agent that analyzes conversations, identifies automation opportunities, and leverages existing capabilities, including Flows and Apex, to drive further automation.
Agentforce Contact Center is enjoying solid momentum, with 1,200 new customers since its March 2026 launch. The company has a large installed base already using Service Cloud and Omni-Channel Routing for digital channels, and flipping the switch on voice interactions seems compelling to many.
The Sales keynote focused on demonstrating Hunter and showing Momentum in action for automated conversation capture, insight extraction, CRM updates, and task creation.
Mark Benioff is adamant that enterprises want to consume AI primarily through packaged software rather than building their own solutions.
Salesforce sees a major opportunity in the 1 billion knowledge workers globally, with only 50 million using Salesforce today. It believes AI could potentially double software spend per knowledge worker by 2030, positioning knowledge workers as the foundation of a new software category. In this context, AIforce and Headless create new ways to bring Salesforce to more users and expand its footprint.
Headless is also an important monetization vector. Salesforce introduced a $50 Headless add-on package earlier this month, with full pricing details expected in November. It has also indicated that access to the Customer 360 Platform, whether through MCPs or APIs, will incur additional Flex Credits based on actual consumption.
Salesforce continues to move and adjust at remarkable speed, impressive for a $50B company.