Dreamforce 2026 Highlights: Top 10 Announcements You Should Know About

Dreamforce 2026 just wrapped up in San Francisco, and if one word had to capture the energy in that room, it would be this: transformation. Not the buzzword kind. The kind where you walk out of Moscone knowing the way your team works inside Salesforce is never going to look exactly the same again. 

Marc Benioff and his team took the stage with a clear message — AI is not replacing the tools businesses rely on, it’s replacing the friction that slows those tools down. With 50,000 attendees on the ground, millions streaming on Salesforce+, and some of the biggest names in tech joining the keynote, Dreamforce 2026 set a new bar for what enterprise AI actually looks like in practice. 

Here are the top 10 announcements from this year’s event that every Salesforce professional, business leader, and CRM user needs to know about. 

Top 10 Announcements from Dreamforce 2026 

1. AIforce: The Biggest Announcement of the Year 

The headline announcement was AIforce — Salesforce’s new interface layer that brings your CRM data, workflows, and permissions into whatever environment your team already works in. That means Claude, Slack, Lightning, Amazon Quick, and beyond. 

The idea is straightforward but significant: instead of your team logging into Salesforce to do their work, Salesforce comes to them. Users who have never opened the Salesforce UI can now update records, trigger workflows, and query data without leaving the environment they prefer. 

AIforce packages four components together: 

• Claudeforce — the extended Salesforce and Anthropic partnership 

• Slackforce — Salesforce context and visualizations built into Slack 

• Agentforce Coworker — an AI teammate embedded directly in the Lightning interface 

• The Headless Toolkit — MCPs, APIs, plugins, and developer tools for builders 

Benioff described it plainly: AI is replacing the UI. The CRM is still the system of record. But the interface your team uses to access it is now up to them. 

2. Claudeforce: Open Beta, Available Now 

If you have been watching the Salesforce and Anthropic partnership develop over the past year, Dreamforce 2026 was the moment it went live for everyone. 

Patrick Stokes, Salesforce President of Applications and Marketing, demonstrated what he called a fully personalized Salesforce interface — one he built himself by having a conversation with Claude. It included a digital version of Salesforce co-founder Parker Harris acting as an assistant in the corner of the screen. 

Claudeforce is now in open beta. Any Salesforce customer can access it through the AppExchange today — search for the Salesforce in Claude plugin, fill out the form with your org ID, and Salesforce will activate it. The plugin comes with 37 prebuilt sales skills and zero data retention, meaning your business data answers questions but never feeds into model training. 

3. Koa: Salesforce’s First CRM Reasoning Model 

Built in partnership with NVIDIA, Koa is Salesforce’s first reasoning model designed specifically for CRM workflows. It was post-trained on NVIDIA Nemotron 3 Super using a synthetic dataset modeled on 27 years of Salesforce CRM intelligence — covering lead generation, opportunity management, and service case resolution. 

Koa is not just another general-purpose AI model repurposed for CRM. It is purpose-built to handle complex, multistep enterprise tasks, and Salesforce claims it produces three times fewer errors than comparable models on CRM-specific tasks. 

Salesforce is already using Koa internally, including a Slack-based agent that helps employees retrieve information and complete everyday tasks. Customer pilots begin in October 2026, with general availability expected in U.S. regions in winter 2026. Early pilot customers include 1-800Accountant, Baxter Credit Union, Formula 1, UChicago Medicine, and Xero. 

4. Agentforce Coworker: Your AI Teammate Inside Lightning 

Agentforce Coworker brings AI reasoning directly into the Salesforce Lightning interface — not through a separate app or external tool, but right inside the environment Salesforce teams already use every day. 

It can reason across accounts, surface historical insights, pull in activity data, and take actions — all while operating within your existing Salesforce permissions and business rules. No new governance headaches. No separate access management. 

Adecco has already rolled Agentforce Coworker out across 40-plus countries following successful pilots in the UK and France, which gives a clear signal of how enterprise deployments can scale. 

5. Slackforce: CRM Context Where Conversations Already Happen 

Slackforce brings Salesforce data and visualizations into Slack without requiring users to switch tabs or hunt through records. Deals, pipelines, account activity — surfaced in the channel where your team is already talking about them. 

This is an extension of what Salesforce started building with Slack after the 2021 acquisition. Dreamforce 2026 made it clear that the vision is fully live now: Slack is not a messaging tool that plugs into Salesforce, it is one of the primary surfaces through which Salesforce is meant to be experienced. 

6. Zero Data Retention: Salesforce’s Commitment to Trust 

Benioff was direct and emphatic about this one. He addressed it plainly during the keynote: when your team works inside Salesforce apps or uses Salesforce platforms, your data does not go into the models. It is not used to train anything. It is not retained by model providers. 

Salesforce says this has been audited, tested, and verified across every part of the AIforce architecture. The zero data retention policy applies across Claudeforce, Koa, Agentforce Coworker, and the Headless Toolkit. 

For companies in regulated industries — financial services, healthcare, legal — this is not a minor detail. It is a foundational requirement. And Salesforce made sure to say it clearly in front of 50,000 people. 

7. Expanded Cloud Partnerships: Google and AWS 

Dreamforce 2026 did not just announce a deepened Anthropic relationship. Salesforce also formalized expanded integrations with both Google Cloud and AWS on the same day. 

On the Google side, Agentforce and Gemini Enterprise are now connected over MCP, with Hyperforce available on Google Cloud. On the AWS side, Salesforce context flows into Amazon Quick, AWS agents can run inside Slack, Agentforce model selection is available through Amazon Bedrock, and Agentforce Voice now works with Amazon Connect. 

The fact that Salesforce made both announcements simultaneously is intentional. It signals that AIforce is being built as a truly model-agnostic layer — your team can choose the AI model that fits, and Salesforce will connect it to your data. 

8. The Jensen Huang Conversation: AI and the ‘No New Regulation’ Moment 

Jensen Huang, CEO of NVIDIA, joined Benioff onstage in a session that was equal parts business strategy and stand-up comedy. But there was substance underneath the back-and-forth. 

Huang was direct: ‘The end of software is nonsense.’ He argued that AI is a layer on top of software, not a replacement for it. He also pushed back on the idea that speed and safety are in tension, calling it a ‘false choice’ and saying companies should pace themselves until they are confident in what they are releasing. 

On regulation, Huang’s position was laissez-faire: market forces are already there, and no new laws are needed. That view will not land the same way in every boardroom, but it reflects where a significant part of Silicon Valley sits heading into 2027. 

9. Claudeforce Keynote with Dario Amodei 

Anthropic co-founder Dario Amodei joined Benioff to speak about the pace of AI development and what it means for businesses trying to keep up. 

Amodei’s observation was striking: even if the technology froze in place today — which it is not doing — most organizations are still only using five to ten percent of the potential value available to them. The technology is moving faster than adoption. 

He gave the example of an Anthropic executive who uses Claude with Salesforce data to pull insights fluently, in real time. Most companies have not yet built workflows that get close to that. For Salesforce customers, this is the real call to action from Dreamforce 2026: the gap between what AI can do and what your team is actually using it for is probably larger than you think. 

10. $27 Million Education Commitment and Sustainability 

Dreamforce is always a platform for Salesforce’s broader commitments, and 2026 was no exception. Benioff announced $27 million in donations to local San Francisco public schools before the keynote officially began. 

Salesforce also reinforced its sustainability commitments. The company has maintained its Net Zero Cloud as a tool to help other organizations meet their own sustainability goals, and Benioff reiterated Salesforce’s long-standing pledge to be a company where trust, equality, and environmental responsibility sit alongside innovation. 

What Dreamforce 2026 Means for Your Salesforce Strategy 

Look across all ten announcements and one theme runs through every single one: Salesforce is becoming the intelligence layer behind your business, not just the application your team logs into. 

AIforce, Claudeforce, Koa, and Agentforce Coworker are all pieces of the same vision. Your CRM data, your workflows, your business rules — made accessible wherever your team works, powered by AI that reasons rather than just retrieves. 

For businesses already running on Salesforce, the opportunity is significant. But navigating what to adopt first, how to sequence implementation, and how to ensure governance keeps up with new capabilities — that takes real expertise. 

As a Salesforce Summit Partner, 360 Degree Cloud works with businesses at exactly this moment. When the platform shifts, we help teams move with it — not scramble to catch up six months later.