Salesforce has announced Agentforce 3: Agentforce’s latest version. This is the first complete management solution of the Life cycle of AI agents to create and optimize autonomous agents prepared for the company. Since its launch in October 2024, Agentforce It has helped thousands of customers to obtain undeniable value thanks to the rise of AI agents, which has meant a saving of more than 50 million dollars for Salesforce, has reduced by 15% the average time of processing of the cases of Engine customers, generated 250,000 dollars in new income to get in just two weeks and increased retention of subscribers of the Globo group by 22%.
With the appearance of a new hybrid workforce, in which human beings and AI agents work side by side, customers demand tools that allow them to continue growing. They need a new way to develop faster, connect agents with reliable workflows, supervise the activity and continually optimize the performance of all equipment.
Agentforce 3 responds to this need with agentforce studiothe first unified experience to design, test, implement and optimize large -scale AI agents. This version also introduces safe support for open interoperability standards such as the models context protocol (MCP), the platform AGENTEXCHANGE Extended with more than 20 pre-validity MCP servers and a faster amortization time thanks to new predefined industrial actions with flexible prices and global availability. All driven by an improved Atlas architecture for higher speed, trust and scalability.
Business Impact
IA agents are no longer experimental, they are becoming generalized, although their scalability remains a challenge. According to the Slack Workflow index, the use of AI agents has increased 233% in six months. On the other hand, 96% of workers say that AI helps them complete tasks that they could not do before. However, most agent platforms lack the tools, governance and observability to climb beyond concept tests. Agentforce 3 changes this paradigm by offering full life cycle tools, safe interoperability and business level controls that organizations need to convert the speed of agents into their competitive advantage.
Agentforce Command Center: How to supervise, measure and optimize AI agents
As IA agents assume more tasks and work together with humans, a new observability layer designed for a new hybrid workforce is needed. Agentforce Command Center is that layer: a complete observability solution that offers leaders a unified panel to supervise the status of agents, measure performance and optimize results. Integrated in Agentforce Studio, completes the life cycle of agents with powerful tools to accelerate their large -scale improvement.
All new agentforce 3 capabilities are based on the Improved Atlas Trust and Reasoning engine
AGENTEXCHANGE EXTENSION: The leading ecosystem of AI agents
As organizations incorporate AI agents into their teams, AGENTEXCHANGE facilitates their equipment with reliable tools They need to generate value quickly. This version expands the market with more than 20 previously validated MCP launch partners. All products included have been subjected to safety reviews and can be implemented without personalized code, so the equipment can improve their agentforce with just clicks. Some of these examples are:
- Box: Agentforce helps content operations to recover the last signed confidentiality agreement of a box MCP service, summarize the key terms and share a Slack update with the accounts team, including the document and the following steps.
- PayPal: Agentforce allows service equipment to manage invoices generation through the PayPal MCP server and agent trade capabilities that are already available in a general way, which allows agentforce to process transactions, create compliant invoices and notify customers by email with attached invoices, all without manual work.
- Stripe: Agentforce allows the financial department to use the STRIP MCP server to issue partial credits, validate the eligibility of reimbursements and publish a conciliation summary in the ERP system.
Optimized agentforce architecture
All new capabilities of Agentforce 3 They are based on the engine Atlas Trust and Reasoning Improved, which provides a base prepared for the company, with less latency, greater precision, global availability and additional control options through LLM housed in the Salesforce infrastructure. With this, Salesforce customers will obtain, among other things, better performance and faster streaming response, greater variety of LLM with Anthropic hosted, greater precision thanks to the web search and online appointments or a greater geographical presence with the incorporation of more languages.
Agentforce 3 It also introduces a simplified and flexible price structure with new sku of Agentforce For sales, services and industrial cloud with prices per user and unlimited use of shares for agents who deal with employees, in order to help equipment quickly and climb their business.
