When 50,000 participants arrive at Salesforce DreamForce conference this week, the enterprise software giant is making its most aggressive bet on artificial intelligence agents yet, positioning itself as the antidote to what it calls an industry-wide “pilot purgatory” in which 95% of enterprise AI projects never make it to production.
The company launched on Monday Agentforce 360a major overhaul of the company’s entire product portfolio, designed to transform enterprises into so-called “agent enterprises” – organizations where artificial intelligence agents work alongside humans to perform up to 40% of sales, service, marketing and operations work.
“We truly live in the era of agentic AI, and I think this is probably the biggest revolution and the biggest technological change that I have ever experienced in my career,” Parker Harris, co-founder and chief technology officer of Salesforce, said during a recent press conference. “In the future, 40% of Fortune 1000 work will likely be done by AI, and humans and AI will actually work together.”
The announcement comes at a pivotal time for Salesforce, which has deployed more than 12,000 AI agents over the past year, building what Harris called a “$7 billion business” around its AI platform. But the launch also comes amid extraordinary turbulence facing CEO Marc Benioff severe reaction for the last comments supporting President Trump and suggesting that National Guard troops should patrol the streets of San Francisco.
Why 95% of enterprise AI projects never launch
The stakes are huge. While companies rushed to experiment with artificial intelligence after ChatGPT emerged two years ago, most enterprise implementations stalled before they reached production, he says. latest MIT research which Salesforce managers cited at length.
“Customers have invested heavily in AI, but they are not getting the benefits,” said Srini Tallapragada, Salesforce president and chief engineering and customer success officer. “95% of enterprise AI pilots fail before production. It’s not for lack of intention. People want to do it. Everyone understands the power of this technology. But why is it so hard?”
According to Tallapragada, the answer is that AI tools remain disconnected from enterprise workflows, data and management systems. “You write hints and hints, and you get frustrated because there’s no context,” he said, describing what he called an “instant doom loop.”
The Salesforce solution is a deeply integrated platform that combines what we call the four components: the Agentforce 360 agent platform, Data 360 for unified data access, Customer 360 applications for business logic, and Slack as the “conversational interface” where people and agents collaborate.
Slack is becoming the front door to Salesforce
Perhaps the most significant strategic change is leveling up Loose — acquired by Salesforce in 2019 for $27.7 billion — as the primary interface to Salesforce itself. The company is effectively transforming its conventional Lightning interface around Slack channels, where sales offers, service requests and data insights will appear as conversations rather than through forms and dashboards.
“Imagine that maybe you don’t log into Salesforce, you don’t see Salesforce, but it’s there. It’s available in Slack because that’s where you do your work,” Harris explained.
The strategy includes embedding Salesforce Agentforce agents for sales, IT services, HR services and analytics directly into Slack, along with a completely rebuilt Slackbot that acts as a personal AI companion. The company is also introducing “Channel expert”, an always-active agent that provides instant responses from conversations on the channel.
To allow third-party AI tools to access Slack conversational data, Salesforce provides an app Real-time search API AND Model context protocol server. Partners including OpenAI, Anthropic, Google, Perplexity, Writer, Dropbox, Notion, and Cursor are creating agents that will run natively in Slack.
“The best way to see the power of a platform is to see the applications and AI agents that are already in development,” Salesforce executive Rob Seaman said during a technical briefing, citing examples of startups that “get tens of thousands of customers who have it installed in 120 days or less.”
Voice and IT services target new markets
In addition to integration with Slack, Salesforce announced a significant expansion of voice interactions and employee support. Agent’s voicenow freely available, transforms traditional IVR systems into natural conversations that can update CRM records, trigger workflows, and seamlessly hand them off to human agents.
IT services offerings represent the most immediate challenge for Salesforce ServiceNowmarket leader. Mudhu Sudhakar, who joined Salesforce two months ago as senior vice president of IT and HR services, pitched the product as a fundamental new approach to employee support.
“Legacy IT service management relies heavily on portals, forms, tickets and manual processes,” said Sudhakar. “We had a few key tenets: conversation first, agent first, and we really focused on providing a conversation for people asking for support and for people providing support.”
The IT services platform includes over 25 dedicated Salesforce agents and over 100 out-of-the-box workflows and connectors that handle everything from password resets to complex incident management.
Early customers report dramatic increases in performance
Customer results suggest this approach is gaining popularity. Reddit Average resolution time reduced from 8.9 minutes to 1.4 minutes, an 84% improvement, while redirecting 46% of cases entirely to AI agents. “This efficiency has allowed us to provide on-demand assistance for complex tasks and increase advertiser satisfaction by 20%,” John Thompson, Reddit’s vice president of sales strategy and operations, said in a statement.
Enginetravel management company, reduced average turnaround time by 15%, saving over $2 million annually. Open the table solved 70% of queries regarding restaurants and bars on their own. And 1-800Accountant achieved a 90% case rejection rate during the critical tax week.
Internal Salesforce implementations may be the most telling. Customer service organization Tallapragada currently handles 1.8 million AI-powered calls per week, with data published on the website help.salesforce.com showing how many agents are responding compared to escalating to humans.
More importantly, Salesforce has deployed AI-enabled sales development representatives to follow up on leads that have not been contacted before due to cost constraints. “Now Agentforce has an SDR that monitors thousands of leads,” Tallapragada explained. The company also increased proactive customer contact by 40%, shifting staff away from reactive support.
A trust layer problem that enterprises cannot ignore
Given enterprise concerns about AI reliability, Salesforce has made significant investments in what it calls its “trust layer” — audit trails, compliance checks, and observability tools that allow organizations to monitor agent behavior at scale.
“You should think of the agent as a human. A digital employee. You need to manage performance like a human. You also need audit trails,” Tallapragada explained.
The company faced this challenge first-hand when scaling the deployment of its own agents. “When we started at Agentforce on Salesforce, we tracked every single message, which is great, all the way down to 1,000, 3,000,” Tallapragada said. “After a million conversations, no man, we can’t do this.”
The platform now includes “Agentforce grid” to sift through millions of conversations to identify and fix problematic patterns. The company also introduced Agent Script, a new scripting language that enables developers to define fine-grained guardrails and deterministic controls on agent behavior.
The data infrastructure will be significantly modernized
At the core of the agent’s capabilities are significant investments in infrastructure. Salesforce Data 360 includes “Intelligent Context” that automatically extracts structured information from unstructured content such as PDFs, diagrams and flowcharts using what the company calls “AI-powered unstructured data pipelines.”
The company also cooperates with Data cubes, dbt laboratoriesAND Snowflake on “Universal Semantic Exchange” – an attempt to standardize the way different platforms define business metrics. The pending $8 billion acquisition of Informatica, which will close soon, will expand metadata management capabilities across the enterprise.
The competitive landscape continues to intensify
Salesforce’s aggressive actions by AI agents come as virtually every major enterprise software vendor employs similar strategies. Microsoft has implemented Copilot across its product line, Google offers agent capabilities through Vertex AI and Gemini, and ServiceNow has launched its own agent offering.
When asked how the Salesforce announcement compares to The latest OpenAI releasesTallapragada emphasized that customers will use multiple AI tools simultaneously. “Most of the time I see they use OpenAI, Gemini, they use Anthropic, like Salesforce, we use all three,” he said.
True differentiation, executives argued, lies not in artificial intelligence models but in integration with business processes and data. Harris framed the competition in terms familiar from Salesforce’s founding: “26 years ago, we just said, let’s make Salesforce automation as effortless as buying a book on Amazon.com. We’re doing the same thing. We want to make agent-based AI as effortless as buying a book on Amazon.”
The company’s customer success stories are impressive, but represent a small fraction of its customer base. With 150,000 Salesforce customers and one million Slack customers, Agentforce’s 12,000 deployments represent about 8% penetration – strong for a one-year-old product line, but hardly ubiquitous.
Company shares down approximately 28% year to date With A relative strength rating of just 15suggests investors remain skeptical. This week’s Dreamforce demonstrations – and the following months of customer deployments – will begin to provide answers to the question of whether Salesforce will finally be able to move enterprise AI from pilot versions to large-scale production, or whether the “$7 billion business” remains more of an aspiration than a reality.