Salesforce’s Seven Named AI Agents: What the September 2026 Agentforce 360 Launch and AI Control Plane Mean for Enterprise Buyers

Salesforce’s September 2026 Agentforce 360 announcement introduces seven named AI agents and a governance layer meant to manage AI systems across an entire enterprise, not just Salesforce’s own products. The launch, timed just before Dreamforce 2026 (September 15–17 in San Francisco), matters beyond Salesforce customers because it reflects where enterprise software is heading: purpose-built agents for specific jobs, plus a control layer to govern them once dozens of agents from different vendors are running inside one company.

This guide is general market analysis, not a product endorsement or implementation recommendation. Evaluate any AI agent platform, including Agentforce, against your own security, compliance and integration requirements.

What did Salesforce announce on September 11, 2026?

Salesforce introduced Agentforce 360, a portfolio of seven named, purpose-built AI agents alongside a new governance and management layer, positioning the release as infrastructure for enterprise AI rather than a single new product.

The announcement arrived days ahead of Dreamforce 2026, Salesforce’s flagship conference running September 15–17 in San Francisco, where the company typically uses the preview window to set the agenda before tens of thousands of attendees and partners arrive.

Who are Salesforce’s seven named AI agents, and what does each one do?

Each of the seven agents is scoped to one business function rather than acting as a general-purpose assistant, which reflects a broader industry shift away from single “do everything” copilots.

• Casey — help and internal service requests
• Paige — IT and HR support
• Carter — shopper and commerce interactions
• Hunter — outbound sales (remains in pilot until November 2026)
• Marshall — supply chain operations
• Piper — inbound sales pipeline
• Fin — customer experience

Most of the seven are generally available at launch; Hunter, the outbound sales agent, is the exception and stays in pilot status until November 2026, suggesting Salesforce wants more validation before it hands a named agent direct control over outbound sales motions.

What is the Trusted Enterprise AI Harness?

The Trusted Enterprise AI Harness is Salesforce’s governance framework for enterprise AI, organized around six pillars: Trusted Context, Trusted Agency, Trusted Action, Trusted Governance, Trusted Security and Trusted Models.

The framework is aimed squarely at a problem enterprise IT and AI strategy teams already have: most large companies are not running one AI agent, they are running many, from multiple vendors, with inconsistent oversight of what each agent can access, act on, and decide. A named set of pillars gives buyers a checklist to evaluate any agent deployment against, not just Salesforce’s own.

What do the six “Trusted” pillars actually cover?

Each of the six pillars maps to a distinct governance question enterprise buyers already ask in security and procurement reviews, which is likely why Salesforce packaged them as a named framework rather than generic marketing language.

• Trusted Context: what data an agent can see, and whether it respects existing data permissions and access boundaries.
• Trusted Agency: what decisions an agent is allowed to make autonomously versus escalate to a human.
• Trusted Action: what systems an agent can actually change or trigger once it decides on a course of action.
• Trusted Governance: the audit trail, approval workflow and policy layer sitting above individual agents.
• Trusted Security: identity, authentication and threat controls applied to non-human agent identities, not just human users.
• Trusted Models: which underlying AI models power an agent, and how their outputs are validated before acting on them.

Framed this way, the Harness reads less like a Salesforce feature list and more like a governance checklist any enterprise could apply when evaluating agents from any vendor — which is a large part of why analysts covering the launch treated it as a platform story rather than a features story.

How does this compare to Microsoft and Google’s enterprise AI governance push?

Salesforce is not alone in shifting the conversation from agent capability to agent governance; Microsoft has pushed its own AI Max and Copilot control tooling, and Google has invested in enterprise-grade oversight for Gemini-based agents, reflecting an industry-wide recognition that ungoverned agents are now the bigger risk than underpowered ones.

What differentiates Salesforce’s pitch is the explicit claim that its Control Plane governs agents built outside Salesforce entirely. Whether that claim holds up under real multi-vendor deployments — where a company might run Salesforce, Microsoft and homegrown agents side by side — is the detail enterprise architects should press vendors on during any evaluation, since a governance layer that only fully works within one ecosystem still leaves the cross-vendor sprawl problem largely unsolved.

What is the AI Control Plane, and why does it matter more than the agents themselves?

The AI Control Plane is the operational layer inside the Harness: it registers agents, sets identity and access policy, manages their lifecycle, evaluates performance, observes behavior, and tracks cost — across Salesforce and third-party AI systems.

This is the more consequential part of the announcement for buyers who are not deep in the Salesforce ecosystem. Named agents are a product story; a control plane that spans third-party AI is an infrastructure story. Industry forecasts cited alongside the announcement suggest close to 40% of enterprise applications will include task-specific AI agents by the end of 2026, up from under 5% in 2025 — a growth curve that makes centralized governance a near-term necessity rather than a future concern.

How does this fit the broader 2026 enterprise AI agent market?

Salesforce’s move follows a pattern already visible across the CRM and enterprise software market: vendors are racing to ship named, task-specific agents while simultaneously admitting that most agent pilots still fail to reach production.

That tension is the real story behind the announcement. Enterprise buyers have spent the past year piloting agentic AI with mixed results, and a governance layer is, in part, an admission that agent proliferation without oversight has created new risk faster than it has created new value. Companies evaluating AI-enabled CRM platforms should weigh governance and interoperability as heavily as the capability of any individual named agent.

What should enterprise technology buyers do with this announcement?

Buyers should treat the Control Plane and Harness as the evaluation criteria, use the pilot status of agents like Hunter as a signal of real-world readiness, and confirm any governance framework covers agents outside its own ecosystem before committing budget.

1. Ask vendors whether their governance tooling manages third-party and custom-built agents, not only their own.
2. Treat “generally available” versus “pilot” status as a proxy for how much production risk a named agent still carries.
3. Map which of the seven functional areas (service, IT/HR, commerce, sales, supply chain, customer experience) actually match unmet needs before evaluating specific agents.
4. Require cost, performance and behavior observability in any agent deployment — the same categories the Control Plane claims to track.
5. Revisit vendor risk assessments for agentic AI deployments already in production, since governance gaps identified now often predate the tooling meant to close them.

Frequently Asked Questions

Are all seven Agentforce agents available now?
Most are generally available as of the September 2026 announcement; Hunter, the outbound sales agent, remains in pilot until November 2026.

Does the AI Control Plane only work with Salesforce products?
No. Salesforce positions it as spanning Salesforce and third-party AI systems, which is central to its pitch as an enterprise-wide governance layer rather than a Salesforce-only tool.

Is this the same as Agentforce 360 announced in 2025?
This September 2026 announcement expands the Agentforce 360 platform with named, function-specific agents and the new governance layer; it builds on rather than replaces the earlier platform.

Why announce this before Dreamforce instead of during it?
Pre-announcing ahead of a flagship conference lets a vendor set the narrative before competing announcements and sessions crowd the news cycle during the event itself.

What is the biggest adoption risk enterprises should watch for?
Agent sprawl without governance — deploying multiple task-specific agents faster than security, compliance and IT teams can oversee what each one can access and do.

Son Güncelleme / Last Updated: September 14, 2026