At Dreamforce 2025, Salesforce didn’t just flirt with AI — it got into a full-blown polyamorous relationship with both OpenAI and Anthropic. Marc Benioff stood on stage, grinning like a man who had just convinced two rival superintelligences to cohabitate peacefully inside his CRM.
The result? GPT-5 and Claude — two of the most advanced large language models on the planet — are now officially part of the Salesforce ecosystem. Einstein (sorry, Agentforce) just got two new brains, and one of them can reason like a philosopher while the other can write haikus about quarterly revenue.
OpenAI + Salesforce: The “Move Fast and Summarize Everything” School of Thought
Salesforce’s expanded partnership with OpenAI is the louder, flashier half of this story. GPT-5 is now embedded across Salesforce products, from the Agentforce platform to Slack and Data Cloud.
Inside ChatGPT, Salesforce’s own Agentforce 360 apps are now available — meaning you can talk to your CRM in natural language. You can literally ask, “Show me my top five opportunities closing this quarter,” and watch ChatGPT conjure the answer, complete with suggested follow-ups or Tableau visualizations. No more report running, just plain English and polite data servitude.
Meanwhile, inside Salesforce itself, OpenAI’s models are selectable options. Developers can build AI copilots powered by GPT-5 that reason, speak, and maybe dream about KPIs. Slack also gets a boost — ChatGPT for Slack now summarizes your endless threads, writes replies, and even codes via Codex, which turns a casual “hey can someone fix this?” into a working patch.
And because no corporate AI rollout is complete without a dash of commerce, there’s Agentic Commerce, where customers can chat with ChatGPT, discover products, and pay instantly via Stripe — all without leaving the conversation. It’s the logical conclusion of a long dream: a salesperson who never sleeps and never stops recommending hiking boots.
Anthropic + Salesforce: The “Keep It Compliant” Contingent
On the other side of the neural coin sits Anthropic’s Claude — Salesforce’s answer to the question, “What if AI, but less chaos?”
Claude is now deeply integrated into Agentforce and Slack, aimed at customers in finance, healthcare, and government — the places where hallucinations are less “quirky” and more “illegal.” Anthropic and Salesforce made a point of emphasizing trust: Claude runs entirely inside Salesforce’s own virtual private cloud, meaning your data never leaves the walled garden.
Financial institutions can now use Claude-powered agents to summarize portfolios, flag regulatory risks, and generate reports that are both useful and auditable. Salesforce engineers, meanwhile, get Claude Code, a programming companion that reads Slack specs and quietly writes documentation without setting anything on fire.
And yes, Claude is also moving into Slack — you can summon it to summarize threads, draft updates, or generate decisions based on conversation context. It’s like a hyper-competent colleague who never needs caffeine and never forgets who approved the pull request.
OpenAI vs. Anthropic: A Tale of Two Neural Philosophies
The contrast between the two integrations is almost poetic.
OpenAI brings speed, flair, and ubiquity — it’s the brand you already talk to at 2 a.m. when you can’t figure out how to write an email. Anthropic brings restraint, compliance, and philosophical calm — the one you’d trust with your tax data or your mother’s medical records.
Salesforce, in its infinite pragmatism, said: Why choose? Now you can pick which AI personality suits your workflow — GPT-5 for creativity and conversation, Claude for governance and grit.
The move also positions Salesforce as the Switzerland of enterprise AI: neutral territory for competing supermodels. OpenAI runs on Azure, Anthropic on AWS, and Salesforce sits serenely in the middle, connecting both through its Einstein Trust Layer and Data Cloud, ensuring everyone plays nice — or at least logs every interaction for audit.
The Agentic Dream (or Corporate Sci-Fi in Beta)
Salesforce’s grand narrative is the Agentic Enterprise — a world where AI agents handle routine work, talk to customers, and collaborate with humans in Slack. The company swears this isn’t about replacing people; it’s about letting you “focus on strategy,” which, in tech translation, means staring thoughtfully at an AI-generated dashboard.
There are real benefits: faster workflows, fewer context switches, and the satisfaction of never having to write “per my last email” again. But there are also real questions: how much trust should we give a probabilistic parrot with access to customer data? And what happens when two AIs disagree in a Slack channel — does the CRM referee?
Still, Dreamforce 2025 made one thing clear: enterprise AI is no longer a feature; it’s the operating system. Salesforce didn’t pick a side in the LLM wars — it picked both, wrapped them in governance, and sold tickets to the show.
In the end, Dreamforce 2025 wasn’t just about Salesforce becoming more “agentic.” It was about watching two of the world’s biggest AIs move into the same digital house — and realizing your CRM might soon be the smartest employee you’ve got.