For Salesforce, from 🤫 hussh
The person-side agent for the agentic enterprise.
We build on Salesforce. It is first-class in our toolchain, and we are proud to be a customer. Agentforce gives every business an agent. 🤫 Private Agent One gives every person one too: their own agent, running on hardware they own, talking to a business's Agentforce agent over MCP, by consent, with a receipt. Your Einstein Trust Layer governs the enterprise's data. Agent One holds the person's consent. Two agents, one trusted handshake.
The other half of the agentic enterprise.
Agentforce is a native MCP client with an enterprise MCP registry. MuleSoft turns any API into a governed MCP server.
🤫 Agent One speaks the same open rails, MCP and A2A. A person's agent talks to your Agentforce agent with no custom wiring.
The person's side of your digital labor vision, connected agent to agent.
The Einstein Trust Layer governs enterprise data with masking, zero retention and secure grounding.
Agent One keeps the person's consent ledger, with a tamper-evident PCHP receipt for every exchange.
A two-sided bridge. You govern the data. The person governs the consent.
Your vision is digital labor at multi-trillion-dollar scale: a billion business agents.
We already run on Salesforce and MuleSoft, and we are building the agent that sits on the person's side.
A foundation in place today, ready for deeper work together.
The system of record, the workflow, and the agent.
Salesforce sells cloud CRM by subscription, with a growing consumption layer for AI through Flex Credits. Its fiscal year ends January 31.
Up 10% year over year, at about a 34% non-GAAP operating margin. The #1 CRM, at scale.
Remaining performance obligations, up 14%. A deep book of committed business.
Up about 205% year over year in Q1 FY27. The agentic enterprise is already real revenue.
A revenue target built on agentic adoption, with digital labor as the engine.
Combined ARR in Q1 FY27. Data and agents, together.
Marc Benioff's view of digital labor: agents working alongside people.
Salesforce holds the customer data, the workflow on top of it, and now the agent layer. It is the ground we already build on.
#1 CRM
The category leader and the system of record for customer relationships across the enterprise.
Data 360
Unifies customer data and federates outside data with zero copy. The governed ground that agents stand on.
Agentforce
The agent platform, now Agentforce 360: a conversational builder, voice, hybrid reasoning and a native MCP client.
MuleSoft
The integration layer that turns any enterprise API into a governed MCP server. The bridge to a person's agent.
Einstein Trust Layer
Secure grounding, data masking and zero data retention with model providers. Governance on the enterprise side.
Slack and the ecosystem
Slack as the place agents work, with Slackbot now an MCP client, and about 6,000 AppExchange apps.
Where Salesforce is going.
The agentic enterprise
Benioff: "agentic AI is the biggest growth opportunity for our customers, and for Salesforce." Digital labor leads.
Data 360 and trust
Governed, unified customer data as the foundation for agents people can trust.
Open interop
MCP across MuleSoft, Agentforce and Slack. Agents that reach any system on governed, standard rails.
Consumption pricing
Flex Credits, so revenue follows agent actions and outcomes.
Digital labor at scale
Business agents working alongside people, at multi-trillion-dollar scale.
Agentforce 360, MCP everywhere, and Slack.
The newest from Salesforce, and the pieces Agent One connects to.
Agentforce 360
Announced at Dreamforce in October 2025 and generally available: Agentforce Builder, Agent Script, Agentforce Voice and Intelligent Context.
MCP across Salesforce
MuleSoft's Anypoint and API Catalog turn any API or Mule app into an MCP server. Agentforce ships a native MCP client and an enterprise MCP Server Registry with allowlists and identity.
Slackbot as an MCP client
Slack orchestrates Agentforce and apps across thousands of integrations. The conversational home of the agentic enterprise.
Flex Credits
Consumption pricing for agent actions. The commercial model a person's consented transactions plug into.
Buy together. Build together. Sell together. Take risk together.
We start from what Salesforce already ships, MCP, MuleSoft and Agentforce, and add the person's side: Agent One and its consent receipt, which we are designing now.
Build together: Agent One and Agentforce over MCP
MuleSoft exposes any system as an MCP server. Agentforce is a native MCP client with an enterprise registry.
🤫 Agent One speaks the same MCP and A2A rails. A person's own agent works with a business's Agentforce agent, with no point-to-point wiring.
Agent-to-agent connection on standards you already ship: the person's side of the agentic enterprise.
Build together: the two-sided consent bridge
The Einstein Trust Layer governs the enterprise's data.
Agent One carries a PCHP receipt for every exchange. The person's consent ledger.
The enterprise governs its data, the person governs their consent, and two agents transact on trust.
Sell together: consent for every Salesforce customer
Salesforce sees a billion business agents doing digital labor.
Agent One adds the person's side, turning a listing into a live, consented, agent-to-agent connection.
A consent layer on top of Salesforce deployments that makes Agentforce more trusted.
Buy together: where we already stand
AppExchange's 6,000 apps and the MuleSoft, Data Cloud and Agentforce stack.
We already run on this stack as a Salesforce and MuleSoft customer, and a consent-first Agent One connector on AppExchange is a natural next step.
Deeper work that starts from a working foundation.
How it fits together.
Two pieces Salesforce ships today, two we are building on top of them.
- 01 · Today
MuleSoft is the bridge
Anypoint and API Catalog turn existing enterprise APIs into governed MCP servers, with security carried forward and no custom code.
- 02 · Today
Agentforce and the MCP registry
Agentforce's native MCP client and enterprise MCP Server Registry, with allowlists, rate limits and identity. The controlled surface a person's Agent One connects to.
- 03 · Next
The person's consent receipt
We are designing Agent One's PCHP handshake on the person's side to pair with the Einstein Trust Layer on the enterprise side.
- 04 · Ahead
Owned and on-device
Agent One running on hardware the person owns, phone, Mac or Puppy One, reaching the enterprise through MCP and A2A.
The people building Salesforce.
Marc Benioff
Chair and CEO (co-founder)
Leads the agentic enterprise and digital labor, which he calls the biggest growth opportunity for Salesforce and its customers.
Robin Washington
President, COO and CFO
Brings the capital discipline and operating rigor behind the FY2030 revenue target.
Srinivas Tallapragada
President, Chief Engineering and Customer Success Officer
Leads the Agentforce, Data Cloud and platform engineering roadmap.
Parker Harris
Co-founder; CTO of Slack
Leads Slack as the conversational surface of the agentic enterprise.
Two agents, one trusted handshake.
MuleSoft bridges the systems. Agentforce's MCP registry is the enterprise's agent surface. The Einstein Trust Layer governs the enterprise's data, and Agent One's receipt governs the person's consent. Two agents, working over open rails, on trust. We build on Salesforce every day, and we would love to build this next part with you. And 🤫 Agent One is free to every American.