Salesforce Closed Fin on Thursday and Put It on the Roster Friday

On Thursday, September 10, Salesforce closed its acquisition of Fin, the AI customer-service company formerly called Intercom, for about $3.6 billion. On Friday, September 11, Fin showed up as one of seven named agents in a new Agentforce lineup, next to Casey, Paige, Carter, Hunter, Marshall, and Piper.

Fitting a $3.6 billion acquisition into a launch the next day took planning. The bigger change is what the launch is selling. Salesforce has spent two years pitching Agentforce as a platform customers use to build their own agents. Now it is also selling agents that come with names, jobs, and published results.

The roster

Salesforce matched each agent to a business function:

• Casey, a customer help agent. Generally available.
• Paige, for employee IT and HR requests. Generally available.
• Carter, a shopping agent. Generally available.
• Marshall, for supply chain. Generally available.
• Piper, for generating inbound sales pipeline. Generally available.
• Fin, a customer agent. Generally available.
• Hunter, for outbound sales. In pilot now, with general availability planned for November 2026.

Salesforce also announced platform features around the agents. Multi-Agent Orchestration is generally available. Agent Optimizer reaches general availability in October. AI Skills in Coworker is in pilot and also reaches general availability in October. Agent Script, which Salesforce calls its "open-source language for agent behavior," lets customers combine AI reasoning with fixed, predictable rules.

The most important piece is Hunter's long-horizon runtime, which lets an agent work toward a goal over days or weeks instead of one conversation. According to Salesforce, "memory preserves context and progress across sessions," "durable execution keeps plans running over time," and "dynamic steering" adjusts behavior based on feedback. SiliconANGLE reports that only Hunter uses this runtime for now.

Why names instead of a toolkit

Named agents are a pricing and packaging choice, not a branding gimmick. A platform needs customers to design the agent, connect the data, write the guardrails, and prove it works, which takes months before any value arrives. A named agent tied to a clear job is closer to hiring someone. The customer can compare it to a role they already staff.

Marc Benioff described this split in June when Salesforce agreed to buy Fin. He said Fin's pre-packaged agents would sit alongside Agentforce's customizable platform, aimed especially at smaller businesses that need to deploy quickly. Friday's launch applies that thinking to the whole product line.

The customer results in the announcement are all framed as work taken off people's plates:

• 50% of Engine's chat inquiries are fully resolved by its help agent, Eva.
• 60% of Perk's sales pipeline is built by Hunter.
• 70% of Autism Queensland's administrative requests are resolved by Paige.
• 90% of core shopper journeys at Hibbett are handled by its AI agent, which went live in six weeks.
• 4x the conversation volume at Asana comes from its website agent, Piper.
• 79% of the conversations Fin sees at Anthropic are resolved without a human.

These are statistics about roles filled, not software features used. That is how a company talks when it wants its agents compared to a headcount budget instead of a software budget.

The numbers behind the push

Salesforce can make this move because Agentforce is already bringing in money. In its fiscal second quarter, reported August 26, Salesforce said Agentforce's annual recurring revenue topped $1.5 billion, up more than 240% from a year earlier. Agentforce and Data 360 together reached nearly $3.9 billion in annual recurring revenue. One quarter earlier, Agentforce alone was at $1.2 billion. Total quarterly revenue was $11.3 billion, up 11%. Salesforce raised its full-year revenue forecast to $46.1 billion to $46.4 billion.

Salesforce also says it has delivered 7 billion Agentic Work Units across Agentforce and Slack, with 3.2 billion in the second quarter alone, up 97% from the quarter before. Readers should treat that metric carefully. Salesforce defines Agentic Work Units and counts them itself, and the number measures activity, not revenue.

Fin closed on a faster timeline than first promised. When the deal was signed in June, Salesforce expected it to close in the fourth quarter of fiscal 2027. By August it had moved the expected close to the third quarter, and the deal closed on September 10. Fin brings more than 30,000 business customers and what the company calls an average 76% resolution rate. It will operate inside Salesforce's AI Labs.

The problems in the lineup

Two customer-service agents. Casey is a customer help agent. Fin is a customer agent. Salesforce's closing announcement says Fin will "complement" Agentforce and that customers can choose between fast deployment on the systems they already use or deeply customized transformation at enterprise scale. That is a reasonable way to divide the market. It also means Salesforce's sales teams now carry two answers to the same question, and one of them has a separate installed base of 30,000 customers that it built without Salesforce's CRM.

The results are chosen by the vendor. Every figure above is a single customer picked by Salesforce. Fin's own average is 76%, and the Anthropic case is above that average. Hibbett's six-week go-live is the story Salesforce wants told about deployment speed. It is not a median.

Hunter is still unproven, and it matters most. Five of the new agents do jobs with short feedback loops, like a ticket or a chat or a request. Hunter works a B2B sales cycle over weeks. Perk's claim that Hunter builds 60% of its pipeline is the boldest number in the announcement, and it comes from an agent that is still in pilot. November's general availability will be the first real public test of whether long-horizon agents hold up at scale.

What to watch

Hunter's November launch, and whether the long-horizon runtime expands to the other six agents.

Whether Fin keeps selling to companies that don't use Salesforce. The closing announcement says Fin "will continue to serve its customers" but doesn't say whether it will stay available to businesses outside the Salesforce CRM.

How the agents are priced. If Salesforce starts charging per named agent instead of by usage, that confirms the headcount comparison and puts it in more direct competition with outsourced service providers.

Third-quarter Agentforce revenue. This will be the first quarter with Fin's revenue included, which will make it harder to separate Agentforce's own growth from what Salesforce acquired.