AI assistants with generic names and no defined roles are giving way to something more structured. On September 11, 2026, Salesforce announced Agentforce 360 — a portfolio of seven named AI agents, each with a specific job title, persistent memory, and in one case, the ability to pursue goals over weeks without being re-triggered by a human.
Seven Agents, Seven Functions
The lineup covers the enterprise stack end to end:
- Casey — customer service across voice, SMS, WhatsApp, and web chat
- Paige — employee IT and HR requests through Slack and internal portals
- Carter — online shopping assistant that compares products and handles checkout in chat
- Marshall — back-office supply chain operations
- Piper — inbound sales lead qualification on websites and in inboxes
- Fin — complex multi-channel support workflows, built on the Intercom platform Salesforce acquired
- Hunter — outbound sales, still in pilot
Six are generally available now. Hunter remains in pilot — and it's the technically most interesting one.
The Long-Horizon Runtime
Hunter is the first Agentforce agent to run on a new long-horizon runtime that lets an agent pursue a goal across days and weeks, not just inside a single chat session. It preserves memory between sessions, maintains a durable execution plan, and adjusts behavior dynamically as circumstances change. For sales outreach, that means a system that can run a multi-week prospecting sequence — following up, adjusting messaging, escalating — without a human re-triggering it after every step. Salesforce describes it as "durable execution with dynamic steering."
The AI Control Plane
Alongside the agents, Salesforce previewed an AI Control Plane designed to govern not just Agentforce agents but third-party AI systems as well. It handles discovery, registration, identity policy, lifecycle management, performance evaluation, and cost control across an organization's entire AI stack. The goal is a single pane of glass: what agents are running, what they're doing, and what they're costing — regardless of vendor.
What the Naming Convention Signals
The decision to give agents names and job titles is deliberate positioning. Salesforce isn't shipping a chatbot that handles everything — it's selling the idea of a structured AI workforce that maps to existing org charts. Casey handles service. Hunter works sales. Each has a job description, not just a system prompt.
Whether that framing holds up operationally is an open question. But as an organizational model for enterprise AI deployment, it's a more coherent vision than a general-purpose assistant doing everything badly. Agentforce 360 bets that enterprises will adopt AI faster when the AI fits recognizable roles rather than requiring them to rethink their structures from scratch.