Since 50,000 participants come to Salesforce Dreamforce Conference This week, the enterprise software giant is making its most aggressive bet yet on artificial intelligence agents, positioning itself as an antidote to what it calls industry-wide “pilot purgatory.” 95% of enterprise AI projects never reach production.
The company launched on Monday Agentforce 360a comprehensive reimagining of its entire product portfolio that goals to remodel corporations into so-called “agent corporations” – organizations during which AI agents work side-by-side with humans to do as much as 40% of the work in sales, service, marketing and operations.
“We're really within the age of agent AI, and I feel it's probably the largest revolution, the largest technological shift that I've ever seen in my profession,” said Parker Harris, co-founder and chief technology officer of Salesforce, during a recent press conference. “In the longer term, 40% of the work in Fortune 1000 corporations will likely be done by AI, and it would be humans and AI actually working together.”
The announcement comes at a pivotal time for Salesforce, which implemented greater than 12,000 AI agent implementations last 12 months while constructing what Harris called a “$7 billion business” around its AI platform. However, the launch also comes amid unusual turmoil faced by CEO Marc Benioff violent backlash for current comments Supporting President Trump and suggested that National Guard troops patrol the streets of San Francisco.
Why 95% of AI projects in corporations never start
There is an unlimited amount at stake. While corporations rushed to experiment with AI after introducing ChatGPT two years ago, most enterprise deployments have stalled before reaching production current MIT research that Salesforce executives quoted at length.
“Customers have invested heavily in AI, but they’re not benefiting from it,” said Srini Tallapragada, president and chief engineering and customer success officer at Salesforce. “95% of enterprise AI pilots fail before production. That's not for lack of intent. People need to do it. Everyone understands the facility of the technology. But why is it so hard?”
The answer, Tallapragada says, lies in keeping AI tools separate from business operations, data and governance systems. “You write prompts, prompts, you get frustrated since the context isn’t there,” he said, describing what he called a “prompt doom loop.”
Salesforce's solution is a deeply integrated platform that connects 4 parts: the Agentforce 360 agent platform, Data 360 for unified data access, Customer 360 apps with business logic, and Slack as a “conversational interface” through which individuals and agents collaborate.
Slack becomes the front door to Salesforce
Perhaps probably the most significant strategic shift is the rise of Loose – Acquired by Salesforce in 2019 for $27.7 billion – as the first interface for Salesforce itself. The company is effectively reimagining its traditional Lightning interface based on Slack channels, where sales deals, service cases and data insights are visible in conversation moderately than through forms and dashboards.
“Imagine perhaps you don't log into Salesforce, you don't see Salesforce, but it surely's there. It involves you in Slack because that's where you do your work,” Harris explained.
The strategy includes embedding Salesforce's Agentforce agents for sales, IT service, HR service and analytics directly in Slack, in addition to a very newly developed Slackbot that acts as a private AI companion. The company can also be launching “Channel expertan always-on agent that delivers quick responses from channel conversations.
To enable third-party AI tools to access Slack conversation data, Salesforce is releasing a Real-time search API And Model Context Protocol server. Partners like OpenAI, Anthropic, Google, Perplexity, Writer, Dropbox, Notion, and Cursor create agents which are natively integrated into Slack.
“The best method to see the facility of the platform is the AI apps and agents which have already been developed,” Rob Seaman, a Salesforce executive, said during a technical briefing, citing examples of startups “which have reached tens of 1000’s of shoppers who installed them in 120 days or less.”
Language and IT services are targeting latest markets
Beyond the Slack integration, Salesforce announced major enhancements to voice-based interactions and worker services. Agentforce voicenow generally available, transforms traditional IVR systems into natural conversations that may update CRM records, trigger workflows and seamlessly hand off to human agents.
IT service offerings represent probably the most immediate challenge for Salesforce ServiceNowthe market leader. Mudhu Sudhakar, who joined Salesforce two months ago as senior vice chairman of IT and HR services, positioned the product as a fundamental reimagining of worker support.
“Legacy IT service management could be very portal, form and ticket focused and is a manual process,” said Sudhakar. “We had just a few core principles: conversation first and agent first, really specializing in making a conversational experience for the people asking for support and the people providing the support.”
The IT services platform includes greater than 25 specialized agents and greater than 100 pre-built workflows and connectors that may handle the whole lot from password resets to complex incident management, based on Salesforce.
The first customers report dramatic increases in efficiency
Customer results suggest the approach is gaining traction. Reddit Average support resolution time was reduced from 8.9 minutes to 1.4 minutes – an 84% improvement – while 46% of cases were fully routed to AI agents. “This efficiency has allowed us to offer on-demand help with complex tasks and increased advertiser satisfaction scores by 20%,” said John Thompson, Reddit's vice chairman of sales strategy and operations, in an announcement.
Motora travel management company, was capable of reduce average processing time by 15%, saving over $2 million annually. OpenTable Solved 70% of restaurant and restaurant inquiries independently. And 1-800Accountant achieved a 90% case dismissal rate throughout the critical tax week.
Salesforce's internal deployments themselves stands out as the most telling. Tallapragada's customer success organization now handles 1.8 million AI-powered conversations weekly, with metrics published at help.salesforce.com Shows what number of agents are responding and what number of agents are escalating as an alternative of escalating to people.
More importantly, Salesforce has deployed AI-powered sales reps to follow up on leads that previously wouldn’t have been contacted because of cost. “Now Agentforce has an SDR that follows up on 1000’s of leads,” Tallapragada explained. The company also increased proactive customer engagement by 40% by shifting employees away from reactive support.
Companies cannot ignore the issue of trust
Given enterprise concerns about AI reliability, Salesforce has invested heavily in what it calls the “trust layer”—audit trails, compliance checks, and observability tools that allow corporations to observe agent behavior at scale.
“You should consider an agent as a human. Digital workforce. You need to manage performance similar to a human. And you wish those audit trails,” Tallapragada explained.
The company experienced this challenge firsthand because it scaled its agent deployment. “When we began at Agentforce at Salesforce, we tracked every message, which is great, as much as 1,000, 3,000,” Tallapragada said. “Once you may have 1,000,000 chats, there’s nobody left, we will’t do it anymore.”
The platform now includes “Agentforce grid“for looking through hundreds of thousands of conversations to discover and resolve problematic patterns. The company also introduced Agent Script, a brand new scripting language that permits developers to define precise guardrails and deterministic controls over agent behavior.
The data infrastructure is being comprehensively modernized
Underlying the agent's capabilities is critical investment in infrastructure. Salesforce Data 360 includes “Intelligent Context,” which robotically extracts structured information from unstructured content equivalent to PDFs, diagrams and flowcharts using what the corporate calls “AI-powered unstructured data pipelines.”
The company also collaborates with Datastones, dbt LabsAnd Snowflake to the “Universal Semantic Interchange,” an try and standardize the best way different platforms define business metrics. Informatica's upcoming $8 billion acquisition, expected to shut soon, will expand metadata management capabilities across the corporate.
The competitive landscape is becoming increasingly intense
Salesforce's aggressive push into AI agents comes at a time when virtually every major enterprise software vendor is pursuing similar strategies. Microsoft has integrated Copilot across its product line, Google offers agent capabilities through Vertex AI and Gemini, and ServiceNow has launched its own agent offerings.
When asked how Salesforce's announcement compares The latest releases from OpenAITallapragada emphasized that customers will use multiple AI tools at the identical time. “Most of the time I see they use OpenAI, they use Gemini, they use Anthropic, similar to Salesforce, we use all three,” he said.
The real difference lies not within the AI models, but in the mixing with business processes and data, executives argued. Harris framed the competition in terms familiar from the founding of Salesforce: “Twenty-six years ago we just said: Let's make Salesforce automation as easy as buying a book on Amazon.com. We're doing the identical thing. We need to make agent AI as easy as buying a book on Amazon.”
The company's customer success stories are impressive, but still represent a small portion of its customer base. With 150,000 Salesforce customers and a million Slack customers, the 12,000 Agentforce deployments represent about 8% penetration – strong for a year-old product line, but hardly ubiquitous.
The company's shares have fallen because the starting of the 12 months around 28% with a Relative strength rating of only 15suggests that investors remain skeptical. This week's Dreamforce demonstrations — and the next months of customer implementations — will provide the primary answers to the query of whether Salesforce can finally move enterprise AI from pilot to full-scale production, or whether the “$7 billion business” stays more wishful pondering than reality.
