Salesforce kicked off its annual Dreamforce customer conference in San Francisco, California, US, on Tuesday, September 15, unveiling a slate of new AI-powered offerings headlined by its first-ever reasoning model, aimed squarely at keeping customers tied to its ecosystem as AI agents reshape enterprise software.The biggest announcement to come out of Dreamforce 2026 was Koa, Salesforce’s first AI reasoning model built on top of Nvidia’s open-weight Nemotron family of LLMs (large language models). Essentially, Koa has been post-trained by Salesforce to make it specifically capable of handling tasks related to sales, marketing, and customer support.Salesforce also launched AIforce, which lets customers carry over enterprise data housed in its products to any AI interface. The enterprise giant further said it is partnering with Air India, enabling the airline to leverage AI agents and other automation capabilities via Salesforce’s Agentforce platform.The company has expanded its Anthropic partnership, including Claudeforce, which allows customers to carry out Salesforce tasks through Claude. It also announced deeper collaborations with Google Cloud and Amazon Web Services (AWS).The rise of increasingly capable AI agents designed to take on more work is forcing SaaS majors like Salesforce to rethink how customers interact with its software and what they are likely to pay for. There is not much evidence that companies are completely abandoning SaaS providers, but reports point to revised contract terms with smaller footprints.Also Read | AI agents are becoming more autonomous. Why researchers are worriedAt the same time, concerns surrounding AI agents seem to have reached a fever pitch over the past few weeks, with Anthropic CEO Dario Amodei calling for an intentional slowdown of frontier AI progress to let AI safety catch up.In an interview with CNBC, Salesforce CEO Marc Benioff said,“We want AI to take care of these things and be held responsible, and take care of these actions and be ethical. The heart of ethics is responsibility.”Story continues below this ad“We really look at a light on our brethren here in San Francisco and all over the world and say, ‘We want you to be ethical in your actions and how you take care of your communities and the world,’ because this technology is very powerful,” he added.Here is an overview of all the announcements from Dreamforce 2026.KoaSalesforce’s first CRM reasoning model, ‘Koa’, is tailored to the enterprise world’s AI needs. Rather than having customers upload files, code, prompts, and feedback directly into third-party AI models, Salesforce is positioning Koa as an open-weight, task-specific alternative.Both Nvidia and Salesforce have reportedly ensured that no actual customer data is used to fine-tune Koa. Instead, the model was post-trained on synthetically generated data that mimics customers’ patterns. It is also designed to follow all of a customer’s data requirements and security embedded within Salesforce.Story continues below this adNotably, Salesforce claimed that Koa’s unique architecture for inference has been designed to be token efficient. This ensures that the model uses fewer tokens to do work tasks that its customers want an AI agent to do. By burning through fewer tokens, using Koa could be cheaper than routing multi-step tasks to Claude or ChatGPT through the Agentforce AI gateway.AIforceAIforce helps customers bring enterprise data, workflows, permissions, security, etc., to any AI interface without having to open the app and navigate through its UI. Users can also generate their own UI and custom live interfaces on the fly within Slack, Claude, and Coworker.Also Read | ‘Alexa+ is more conversational’: Amazon execs on bringing AI-powered assistant to IndiaUnder the hood, AIforce is powered by Salesforce’s Headless Toolkit, an open architecture that exposes every element of the Salesforce platform, and gives developers MCPs, APIs, plug-ins, skills, and tools to create customised AI experiences on top of Salesforce’s architecture.AIforce has been designed keeping in mind the principles of security, governance, and zero data retention, according to Salesforce.ClaudeforceStory continues below this adSalesforce has upgraded its existing Claudeforce offering, rolled out in partnership with Anthropic. It essentially allows companies to use Claude as their AI interface while their data remains in Salesforce’s system of record, secured by its infrastructure.Claudeforce now offers over 37 pre-built sales skills that let sales professionals automate pipeline updates, access revenue context, and take governed action. It is currently rolled out in open beta.Under the partnership, the Tata-owned carrier will use Agentforce-powered AI agents for processing refund requests in under four hours, and resolving baggage and loyalty-related claims. It will also use Agentforce for managing eligibility checks and reissuing tickets with requested name changes in under 30 minutes.The AI-agent-driven transformation of Air India is part of its broader Vihaan AI project.Story continues below this ad“We’ve built incredible teams in India. I think this is a huge opportunity for India. That is, we’re going to rebuild all of our enterprise systems to be AI-first, and I think these intelligent interfaces-they’re going to do some of the work, and then they’re going to need partners, trailblazers, and we have millions and millions of trailblazers in India. So I’m sure that they will be a key part of this incredible story,” Benioff was quoted as saying by PTI.