Introducing AgentExchange, the marketplace for agents and Salesforce applications
Salesforce
Salesforce reorganizes one of the most crowded pieces of its offering and tries to make it a more manageable infrastructure for artificial intelligence in the business. The company has made available AgentExchange, a new unified marketplace that combines AppExchange, Slack Marketplace, and the Agentforce ecosystem. The goal is to place enterprise applications, AI agents, integrations, and Slack tools in the same space, thus simplifying the discovery, purchase, activation, and management of solutions.
A single marketplace for three ecosystems
With AgentExchange, Salesforce brings together over 10,000 Salesforce applications, more than 1,000 agents, sub-agents, tools, and MCP servers from the Agentforce universe, and over 2,600 Slack apps and agents. Until now, these environments were distinct, with different paths to search for, evaluate, purchase, and deploy a solution.
The new marketplace aims to concentrate everything into a single experience, both for those working within Salesforce and those operating in Slack. Salesforce’s idea is to accelerate agentic AI by providing customers with a catalog of components, applications, and integrations that can be adopted at the enterprise level. By further simplifying approval, installation, and configuration processes: with AgentExchange, the time from purchase to actual usage of the solution is almost immediate.
Search becomes semantic and then conversational
Functionally, the new marketplace is based on a semantic search powered by Data 360, a real-time data engine that unifies fragmented information from any source. The idea is to move beyond traditional keyword search and connect the business need more directly with the available solutions. If a user is looking for a way to speed up the management of sales contracts, the engine should interpret the request as a need related, for example, to electronic signature tools or document workflows.
Salesforce also promises contextually built suggestions, taking into account already installed applications, the configuration in use, and existing workflows. From fall 2026, a further step is expected with the introduction of conversational search. In that case, the user will be able to refine the results with follow-up questions and compare solutions in a more dynamic way.
Slack becomes a discovery environment, as well as execution
Another relevant piece concerns Slack. AgentExchange will not only be accessible from the collaboration platform but is designed to work inside the operational flow. The integrated browser will allow the discovery and evaluation of solutions without leaving the workspace. Subsequently, according to the roadmap, Slackbot is expected to add contextual suggestions in natural language, based on the team, existing workflows, and the task at hand.
The same setup applies to Agentforce Builder, the solution for building custom AI agents internally. During the construction of an agent, the platform should suggest relevant components, such as connectors to knowledge bases, case management tools, or sentiment analysis features.
Security and governance remain the decisive filter
Salesforce emphasizes the theme of trust greatly. Each solution is described as being subjected to security checks to meet enterprise standards for security, privacy, and compliance. Administrators maintain control over permissions, data access, API domains, and integration boundaries. They can also define who accesses each solution based on user, role, and profile.
For a company, this is probably the real dividing line. As the number of agents and connectors increases, so does the risk surface area. Therefore, AgentExchange is proposed as an environment that attempts to combine ecosystem openness with centralized control. The presence of reviews, ratings, use cases, and compliance indicators goes in the same direction: reducing uncertainty during the selection phase.
Availability
Salesforce makes AgentExchange available immediately for Salesforce and Slack, the semantic and contextual search based on Data 360, discovery in Agentforce Builder, and tools related to private offerings and go-to-market. But the weight of the announcement is primarily measured by what will arrive in the coming months.
By fall 2026, agentic search, conversational discovery, more personalized recommendations, and intelligent comparison between applications are expected. During the same period, AI-based publishing tools for ISVs and improved listing pages should arrive. In winter 2026, a simplified installer, new contextual discovery features in Agentforce Builder and Slackbot, and checkout with public offers integrated with go-to-market apps are expected.
"Twenty years ago, we first introduced the concept of a marketplace for applications with AppExchange, maintaining a leadership position ever since. Today we evolve this model for the agentic era, bringing together in one environment everything customers need to develop and deploy agents, applications, and integrations on an enterprise scale, with the levels of control they expect. With the new AgentExchange, partners can more effectively access the entire Salesforce installed base and have tools that allow them to develop, manage, and scale distribution more efficiently than ever," comments Brian Landsman, CEO of AgentExchange and EVP Global Partnerships at Salesforce.