Salesforce launches Agentforce 3 with AI agent observability and MCP support


Join a reliable event by enterprise leaders in about two decades. VB Transform, Real Enterprise AI strategy brings together people who build. Learn more


Righteousness By applying the AI ​​Agency Platform facing enterprises with the largest obstacles, which are the largest obstacles in the scale, by contacting the AI ​​Agent Platform:

Company Agentforce 3 The release presents a comprehensive acquaintance “Command center“This gives AI agent performance, AI agent performance real-time viewing, hundreds of agents meet hundreds of external work tools in line with interaction standards that allow you to contact hundreds of external work tools.

Timing reflects the requirement of demand for AI agents. According to Salesforce, the use of AI agent has jumped 233% with more than 8,000 customers to place technology in six months. Early adopts see measurable income: Engine Customer box handling time 15%, whereas 1-800ANT 70% of administrative conversation inquiries were obtained during the peak tax season.

“If we do not have thousands, we have hundreds of live applications and work on a scale,” he said. The company acted strongly than experimental deployments, “AI agents are no longer an experimental. Indeed, they really moved deep into the part of the enterprise.”

“We listened to our customers in the last few months and continued to have a rapid technology innovation,” said Adam Evans, the Executive Assistant and Salesforce AI, announced the improvement of the platform during a live event. “The result is a key leap for our platform, which brings more confidence and reporting to the agentforce 3, higher intelligence, higher performance and placement of each agent force.”

Global food giant Pepsiko, the enterprise pursues the AI ​​Agent Revolution

Among companies covering this technology PepsicoThe agency, which placed an agent force as part of the more widespread EU’s global operations. In an exclusive interview with VentureBeat, Athena Kanioura, PepsiCo’s head strategy and transformation officer, the company’s evolution in a market that increases the placement is very important.

“As a long time partner, Salesforce, especially because the customer scenery is developing, the better complexity and increase our trade is” Kaniooooooooooooooooooooooooooooooo.

The food and beverage giant consumed a billion times a billion times worldwide is important to manage systems and simplifying processes, “wherever they are in the places where they are”.

Pepsico’s seven-year connection with Salesforce, the company placed the company to move to AI agents quickly. “Agentforforce is excited to see how we can develop daily experience for our field vendors and explore and delete easy concepts in real time,” he said.

Missing fabric: Why enterprise AI needs monitoring and control in real time

This Command center The company represents the response to Salesforce to a critical cavity in the AI ​​market. Companies are in a hurry to place AI agents for customer service, sales and operational assignments, many do not have visibility to how these digital workers fulfilled or affect the work results.

Govindarajan described the establishments facing the facilities moving outside the pilot programs: “The EU agent is something to build a demo, but there is a different standard when you build an agent system.” Explained that the AI ​​needs to understand when the agents are fighting and when they will bring people to work flow.

“Teams can’t see what agents do – or are fast enough,” the company admitted to the announcement. The new observation platform provides detailed analytics on recommendations working in the EU for health monitoring and optimization with agent interactions, health monitoring, real-time signals.

The system, Govindarajan’s “two problems” – operational problems that occur after initial placement. “You can be very agent for many people and see how it affects the task that needs to be done on a scale,” he said. This includes the management of works between digital agents and human employees, when complex decisions or approvals are required.

The system seizes all agent activities in the sales area Information using Outlying standard that provides integration with existing monitoring tools such as Datadog and other enterprise systems. It is necessary to monitor the AI ​​Agency for enterprises in the AI ​​Agency.

Open Standards and Safe Integration: How AI agents have joined enterprise systems

Perhaps a more important thing seller hugs Model context protocol (MCP), open standard for the interaction of the AI ​​agent. The platform includes local MCP support, native MCP support that allows agent reinforcing agents to connect with any MCP compatible server without special development work.

“There is a common interaction and then there is something we have in business-class interaction.” “If it’s not an enterprise rate, this is like a spark interope.” The main difference, he said there are management and control mechanisms required by the enterprise.

Extended AgentExchange Marketplace allows you to access pre-built integration with more than 30 partners including enterprises Amazon Web Services, Boxing, Google Cloud, Ibm, Paypaland Wale. Lerhaupt said the company’s “25 plus” with Vetted MCP servers of the company, “25 plus”, which offers invoice capabilities and boxes to obtain a document via MCP applications, said he started with “25 plus” vetted MCP servers.

“In a world filled with AI tools, AgentForce said, not only for the first technology, our technology ecosystem, our work and aI strategy, our standards and frames, our standards and frameworks,” Kanioura said.

Performance increasing: Faster AI models and advanced security for adjustable industry

Sellers under new features called something developed “Atlas“An enterprise-class performance and architecture for security. The platform delays 50% lower than in January 2025, the flow of reaction to real-time user experience and automatic failures between AI Model Providers and AI Model Providers

For adjustable industries, AI models apply to critical security concerns to host AI models within its infrastructure. “With anthropic, all stacks will work within the sale of the sales lelf,” said Govindaraj. “Calls do not go to Openai and the traffic sale will work within the VPC. This is what we work for an adjustable industry.”

Critical for adjustable industry, Salesforce now hosts Anthropik’s clod models through direct infrastructure Amazon BedrokSustaining sensitive data in the Salesforce security perimeter. The company plans to add Google’s Gemini Models The end of this year, enterprises make more choices for AI model management.

Expands the global availability for Brazil, in England, India, Japan, Japanese, Japanese, Japanese, Japanese, Japanese, Japanese, Japanese, Japanese, Japanese, Japanese, Japanese, Japanese, Japanese, Japanese, Japanese, Japanese, Japanese, Japanese, Japanese, Japanese, Japanese, Japanese, Japanese, Japanese, Japanese, Japanese, Japanese, Japanese, Japanese, Japanese, Japanese, Japanese, Japanese, Japanese, Japanese, Japanese, Japanese, Japanese, Japanese, Japanese, Japanese, Japanese, Japanese, Japanese, Japanese, Japanese, Japanese, Japanese, Japanese, Japanese, Japanese, Japanese, Japanese, Japanese, Japanese, Japanese and Portuguese.

From Zero AI Agent: How to speed in pre-established industrial actions

It was established on Salesforce to accept that enterprises need faster return to AI investments 200 pre-adjustable industrial campaign – With more than 100, this summer was added only. These patients are designed to apply AI agents that are designed to help the offspring of advertising proposals in the media, developing companies, more than zero.

The results demonstrate the maturity of the platform. During the tax season, 70% of the 70% defectivity of 1-800Acountmant said: “73% of the entire restaurant managed by OpenTable Agents,” and “using WhatsApp, the Colombian customer service operation only reduced a 71% reduction in three weeks.”

The company also presented more flexible prices, including the real AI case of employees facing employees and a simple AI case than simple talks.

New Digital Workforce: What enterprise means for an EI adoption work operations

The company has increasingly noticing AI agents, not a simple automation tool, as digital workers, the stakes have not been high. Companies that successfully scale AI agents are developing significantly while fighting for the risk of management and control, to gain significant competitive advantages.

Govindarajan sees the main changes in how to organize the case: “New roles are created for people who manage the agent,” he said. “A CIO may ask:” I have seven agents in my enterprise, are there a wide place? “But someone who works on a particular marketing agent has a different lens in the same problem.”

Looking forward, Lerhaupt, posted the current moment as a transformation: “Your personal computer, then there was internet and now there were many agents.” “Many agent revolutions and agents from the placement of the evolution were called” to connect each other to make agents very complex new jobs. “

The transformation for Pepsico goes beyond the benefits of efficiency. “AI and technology once changed in unimaginable enterprise operations,” Kaniooura said. “The work we do with the power of the agent is an element of a company that paves a stronger and adaptive company as a stronger company of Pepsico.”

Competitive landscape strengthens as large technology companies competitions to create AI agent platforms. Asked about Microsoft, Google, Google and Amazon, Govindarajan’s competition, Salezforce’s integration advantages: “We can watch the entire work period within the enterprise’s ecosystem,” he said. “We can identify streams and interactions in the enterprise and open and expanded when you set up your data, actions, actions and efficient.”

This Agentforce 3 The platform is available on a Monday with several features including the host Anthrop models and the full Command center To pass through August. But perhaps the most open sign of technology’s training is not in the list of features – in the confidence of companies such as Pepsico to bet on the digital transformation in AI agents, which can be measured and managed by Pepsiico.



Source link

Leave a Reply

Your email address will not be published. Required fields are marked *