Salesforce’s Vision for Self-Working Software at Dreamforce 2026

Remember when traditional CRM tools felt like glorified, hyper-expensive phone books?

Sales reps spent hours typing meeting notes, updating deal stages, and copying records between tabs. If you worked in enterprise tech, managing Salesforce software usually meant spending half your afternoon just clicking through endless form screens.

Those manual days are officially ending.

At Salesforce Dreamforce 2026, CEO Marc Benioff made a bold statement: the traditional user interface is fading into the background. Instead of asking employees to manage their tools, modern Salesforce software is transforming into an autonomous, self-working engine that handles complex business tasks on its own.

This isn’t just about sticking a smart chatbot into a sidebar; it’s a ground-up redesign of how enterprise Salesforce software functions across global organizations.

From Co-Pilots to Autonomous Execution

For the last couple of years, enterprise apps were obsessed with AI “co-pilots.” You typed a prompt, and a side panel gave you a neat text answer.

Helpful? Sure. But you still had to do the actual clicking.

The major shift on the Salesforce Agentforce platform focuses on fully autonomous execution. Instead of waiting for manual commands, self-working Salesforce software actively monitors live sales pipelines, schedules follow-up tasks, updates customer profiles, and resolves service tickets behind the scenes.

Agentforce Coworker: A proactive digital teammate that works directly alongside human teams. Instead of clicking through ten separate menus in traditional Salesforce software, employees can let digital coworkers handle administrative upkeep across connected systems.

• Autonomous Action: Tasks execute across connected systems without requiring manual triggers.
• Work Anywhere: Intelligence streams seamlessly into Slack, Claude, or custom dashboards.
• Enterprise Guardrails: Automated choices run inside strict operational boundaries.

The Brain Behind the Engine: Koa AI Reasoning Model

Self-working enterprise tools need a specialized brain to make accurate operational decisions. General-purpose AI models are great at drafting emails, but they often stumble over complex corporate hierarchies and strict compliance rules.

To solve this, Salesforce introduced Koa, a dedicated AI reasoning model built with NVIDIA and trained on the Nemotron framework.

Unlike generic large language models, Koa is built specifically to power Salesforce software workflows. It evaluates sales opportunities, flags revenue risks, and coordinates multi-step actions using real-time corporate data—all while keeping proprietary data strictly secured.

CRM as an Invisible Operating System

The underlying theme at Dreamforce is simple: enterprise databases aren’t going away, but manual data entry certainly is.

By making records and business logic accessible through open tools like AIforce and Slackforce, the platform turns core Salesforce software into an invisible operational backbone.

Whether your team uses a Salesforce AI CRM dashboard or relies on background agents, modern Salesforce software ensures everyone works with accurate, up-to-date information. The ultimate goal? Less manual clicking, fewer process bottlenecks, and more time spent growing the business.

Frequently Asked Questions