Dreamforce 2026 is officially underway, and Day 1 came with a lot to unpack.
Agentforce is still at the center, but the conversation has expanded beyond simply building AI agents.
Salesforce is now talking about where those agents work, which models power them, how they access enterprise context, and how businesses can manage them once they start operating across multiple systems.
From AIforce and Koa to new Agentforce agents, deeper Claude and Slack experiences, and enterprise AI governance, here are the announcements that stood out from Day 1.
Day 1 was about the bigger picture. Salesforce introduced new pieces of its AI stack, showed how Agentforce is evolving, and spent more time on where agents work, what powers them, and how enterprises can manage them at scale.
One of the biggest Day 1 announcements was AIforce, Salesforce’s new interface layer for enterprise AI.
The idea is to make Salesforce data, workflows, business logic, permissions, and actions available inside other AI experiences instead of making users return to CRM for every task.
Claude, Slack, and Agentforce Coworker are some of the first places where this approach is showing up, giving users more ways to work with Salesforce context without constantly switching between tools.
Claude and Slack were both central to the Day 1 story.
Salesforce showed how Claude can work with Salesforce context for tasks like pipeline reviews, prospecting, and account work, while Slackforce brings CRM data and actions directly into Slack.
Salesforce wants people to be able to work with CRM information from the tools they are already using instead of treating Salesforce as the only place where the work happens.
Salesforce also introduced Koa, its first CRM reasoning model developed with NVIDIA. Koa is designed specifically around CRM workflows such as lead qualification, case routing, opportunity management, and multi-step business processes.
Salesforce is still keeping its broader multi-model strategy, so Koa is not replacing Claude, Gemini, or other models. It simply gives customers another model built specifically around Salesforce-style work.
Another noticeable shift is how Salesforce is positioning Agentforce.
Instead of focusing only on tools that help businesses build agents from scratch, Salesforce is introducing more agents designed around specific roles, including:
The goal is to give companies a stronger starting point and then let them configure those agents around their own data and processes.
As Salesforce talks about deploying more agents across the enterprise, it is also putting more attention on governance, security, and control.
Technologies such as Agent Fabric, Salesforce Guardian, and the broader Trusted Enterprise AI Harness are meant to help businesses understand which agents are running, what they can access, how they behave, and whether they are following company policies.
This part of the story may be less flashy than the demos, but it is likely to become much more important as companies move from a few AI pilots to much larger agent deployments.
The customer examples helped make the Day 1 announcements more practical.
Siemens showed how Agentforce can support inbound sales and operational workflows, while Air India shared how it is using Agentforce across service processes such as refunds and passenger name changes.
These examples showed the direction Salesforce is pushing toward: agents that do more than answer questions and are increasingly involved in real business processes across systems.
Another thing that stood out on Day 1 was just how open Salesforce wants its AI strategy to be. Even as it builds more of its own stack with products like Koa, it is still leaning heavily into partnerships with Google Cloud and AWS.
With Google Cloud, Salesforce is connecting its headless architecture with Gemini Enterprise, so agents can work with Salesforce business context across both platforms.
The AWS partnership is expanding too, with more connections across models, agents, enterprise data, and the tools employees already use.
Salesforce is building more of the AI stack itself, but it is still leaving the door open for customers to choose the models and platforms that fit their business.
If Day 1 introduced the bigger pieces of Salesforce’s AI stack, Day 2 showed more of how those pieces work together.
Across the Data 360, Agentforce, SMB & Startup, and AIforce keynotes, the focus moved into context, execution, ROI, and where agents actually fit into everyday work.
The Data 360 keynote focused on giving agents the right business context, not just more data.
Salesforce introduced the Agent Context Engine, which brings together data, memory, semantics, permissions, and business interactions so context can move between agents and humans instead of staying locked inside individual agent experiences.
Data 360 is also becoming more headless, with access through APIs, MCP, and skills, allowing agents to work with trusted Salesforce context from interfaces such as Claude.
A few things stood out:
The Agentforce keynote focused less on what agents can do and more on what happens once they are actually in production.
Salesforce showed an approach built around starting with a focused use case, extending the agent as the use case grows, and continuously measuring and improving it.
The keynote also highlighted Agentforce Skills for reusable business processes, Agentforce Observability for monitoring performance, and long-horizon agents for work that continues over time.
The Fulton Bank demo showed Agentforce Coworker using business context and calling specialized agents when deeper expertise was needed.
The SMB & Startup keynote focused on what agents can look like for smaller teams.
Salesforce showcased Slackbot, Piper, Agentforce Coworker, and Fin across sales, support, research, and recurring business workflows.
One of the more interesting examples came from Sastr.AI, which operates with three human employees and 21 agents in production. Its Amelia AI agent has handled more than three million interactions across its community.
For smaller teams, the pitch is simple: agents can take on specific pieces of work without headcount growing at the same pace.
Salesforce also expanded Missionforce with OpenAI and NVIDIA, bringing more model choice and AI capabilities into government and sensitive environments.
OpenAI models are coming into Missionforce through Salesforce Government Cloud and Amazon Bedrock, while NVIDIA is supporting mission-specific model training and deployment.
Salesforce also introduced the Missionforce Policy Engine, designed to turn approved policies into structured rules and workflows with human review built in.
Live Nation shared how its Agentforce use is moving beyond festivals.
Its Melody festival agent handled more than 37,000 fan interactions and 17,000 customer service sessions during its 12-day launch window.
Now Live Nation is expanding that approach with Venue Agent, which will handle common questions around parking, accessibility, door times, and other event-day information.
Salesforce also showed AIforce Command Center, which brings together business metrics, root-cause analysis, what-if scenarios, and next actions in one experience.
Instead of only showing what changed, the idea is to help teams understand why it changed and what they can do next.
Commerce Cloud also got more attention on Day 2.
Salesforce showed Agentic Commerce experiences designed to understand customer intent instead of relying only on keyword search.
The West Marine example showed how AI-powered product discovery can make the buying journey more relevant from the start.
Day 3 brought the Agentic Enterprise closer to everyday work.
From real Agentforce deployments to Slack, Field Service, and Marketing updates, the focus was simple: where can agents actually take work off people’s plates?
The Customer Success keynote showed what happens when companies move Agentforce beyond pilots and into real operations.
The Grout Guy was one of the standout stories. The Australian field service company is using multiple agents across leads, work orders, quoting, scheduling, and technician workflows.
Work orders that once took up to two days can now be processed in under two minutes.
Other highlights:
Slack is moving beyond messaging and becoming a place where teams can work with Salesforce data, agents, and business tools without constantly switching apps.
One of the biggest announcements was Slackforce Surfaces, which can turn conversations and business data into live dashboards, reports, and other interactive views.
More Slack updates:
Field Service focused on using agents to reduce the manual work around scheduling, dispatch, and technician support.
Scheduling Agent 2.0 was a major update. It can use customer history, asset information, warranties, technician skills, SLAs, and business rules to help find the right appointment.
More Field Service updates:
The Marketing keynote focused on a growing shift: people are increasingly discovering brands through AI answer engines before reaching a website or campaign.
That made Palmata one of the more interesting announcements. It helps marketers understand how AI systems talk about their brand, where those answers come from, and what content could improve them.
More Marketing updates:
Dreamforce 2026 showed how much Salesforce is changing around AI.
Across the three days, the focus kept coming back to better context from Data 360, more capable agents in Agentforce, Slack becoming a bigger part of how work gets done, and AI moving deeper into sales, service, marketing, IT, and field operations.
But announcements are the easy part.
For businesses, the harder part is figuring out what is actually worth using, where it fits, what data it needs, and how it connects with the systems already in place.
At MIDCAI, we help businesses with Agentforce consulting and implementation, Salesforce Data 360, Slack implementation, and the integrations needed to make these systems work together.
Dreamforce brought the shiny new stuff. Let’s build the part your business will actually use.
5+ years of experience in content strategy, marketing, and technical communication. At MIDCAI, I focus on turning ideas around Salesforce, data, and AI into clear, actionable insights that help enterprises attract the right audience and drive real adoption.
Get in touch with us for any enquiries and questions
Define your goals and identify areas where technology can add value to your business
We are looking for passionate people to join us on our mission.
where your skills fuel innovation and your growth powers ours
.jpg)
