Dreamforce 2026 Recap: What AIforce, Koa & Claudeforce Mean for Enterprise Salesforce Buyers

After a year of speculation that Salesforce was pivoting away from CRM into a pure AI and data play, Dreamforce 2026 put that question to rest. The message was clear: Salesforce is going all-in on AI, but not at the expense of the CRM backbone that built its business.

Notably, several products rebranded under the Agentforce name last year, such as Agentforce Sales, have quietly reverted to their original names: Sales Cloud, Service Cloud and so on. AI is now the layer on top, not the replacement.

AIforce is Salesforce's bet on agentic, dynamic interfaces that go beyond the standard Salesforce app experience. Under the hood, a new headless architecture lets teams access the Salesforce platform from whatever agentic tool they are already using.

That is a meaningful shift for organizations that do not want to force every workflow through the traditional Salesforce UI, and it changes how you should think about where user experience actually gets built.

Salesforce introduced Koa, its first CRM-specific reasoning model, built on NVIDIA Nemotron and currently in pilot. This matters because it signals Salesforce wants more control over the reasoning layer that powers Agentforce decisions, rather than depending entirely on third-party models.

Perhaps the most consequential announcement for day-to-day sales teams: Salesforce and Anthropic unveiled an expanded partnership pairing Claude's reasoning with Salesforce's enterprise data, workflows and governance.

The rollout starts with Salesforce in Claude, currently in beta, giving sales teams direct access to CRM data and actions such as meeting prep and pipeline review from inside Claude itself. For buyers hesitant about handing sensitive CRM data to an AI layer, Salesforce has been explicit that it will not let LLM vendors train on customer data, keeping everything inside the governed Salesforce ecosystem.

Rather than asking customers to imagine every possible use case, Salesforce announced seven purpose-built agents with stated availability. Six of the seven are available now, a notable shift from last year's mostly-demo posture.

The clearest signal that Agentforce has moved past the hype cycle: this year's keynotes leaned heavily on customer results instead of new feature reveals. Fulton Bank reported saving 80,000 hours through Agentforce, while attributing $389 million in loans and deposits to the technology and cutting complaint resolution times by four days.

The Adecco Group announced an Agentforce Coworker rollout spanning more than 40 countries, starting with a pilot in the UK and France.

If you are evaluating whether to invest in Agentforce this year, the calculus has changed. The open questions are less about whether the technology works and more about which agent to start with and how your data and permissions hold up under it.

Evaluating six new agents, a new interface layer and a fresh reasoning model is a lot to map against your own Salesforce org, especially if you are running a heavily customized instance.

AX3 works with enterprise teams to scope, implement and govern Agentforce rollouts, from picking the right first agent to making sure your data and permissions are ready for it. Talk to AX3 about your Agentforce roadmap.