What Salesforce's Own Newsroom Post Actually Says
Published: September 17, 2026 | Category: AI, Business | By Mahesh | Source: Salesforce's official newsroom announcement, primary data current as of September 17, 2026
Today's new product is a portfolio of seven named AI agents that Salesforce introduced on September 11, 2026, ahead of its annual Dreamforce conference. Salesforce's own newsroom post opens with a line worth sitting with: "the hardest part is no longer getting AI to answer a question. It's getting AI to do the work that matters across every business function."[1] Depth Grid has tracked the broader shift from raw model capability toward packaged, workflow-embedded agents in OpenAI Will Now Rent You the Same Agent Infrastructure That Runs Codex; this launch pushes that same shift one step further, selling not infrastructure for building agents, but agents that already have a job description.
What Actually Launched, Agent by Agent
Claim: Salesforce packaged seven agents around specific business functions rather than releasing a single general-purpose assistant. Source: Salesforce's own post lists each by name and role: "Casey, your help agent, resolves customer service issues across voice, SMS, WhatsApp, and web chat, with pre-built support for FAQs, returns, account management, human escalation, and more"; "Paige, your IT & HR service agent, resolves requests across Slack, portals, and the tools employees already use"; "Carter, your shopper agent, helps shoppers discover and compare products, get questions answered, and convert with in-chat checkout"; "Marshall, your supply chain agent, orchestrates end-to-end back-office processes, automates manual work with deterministic execution, and provides an audit record of every action"; "Piper, your inbound pipe gen agent, works across websites and inboxes to engage, qualify, and convert inbound leads into pipeline for B2B sales and marketing teams"; and "Fin, your customer agent, resolves complex customer experience workflows across every channel," built on Operator and Fin Apex, technology Salesforce gained through its acquisition of the company formerly known as Intercom.[1][2] All six of these reached general availability the same day.[1] Analysis: The consistent design choice across all six agents is that each one is scoped narrowly to a single job function rather than positioned as a flexible, do-anything assistant, which mirrors how Salesforce has historically sold its core CRM software as role-based modules (Sales Cloud, Service Cloud, Commerce Cloud) rather than one undifferentiated product. Packaging agents the same way is a deliberate continuation of a go-to-market model Salesforce already knows works for enterprise buyers who budget by department, not by technology category. Published: September 11, 2026.
Hunter, the One Exception, and Why It Matters
Claim: The seventh agent, an outbound sales agent named Hunter, is deliberately held back in pilot because it runs on fundamentally different, more experimental technology than the other six. Source: Salesforce's post states Hunter is in "pilot now; GA November '26," and explains why: "Salesforce built a new long-horizon runtime for Agentforce, enabling agents to pursue goals across days and weeks instead of completing only a task or interaction," adding that "Hunter is the first agent to run on the long-horizon runtime," built on three underlying capabilities: "memory" that "preserves context and progress across sessions," "durable execution" that "keeps plans running over time and allows an agent to resume or course-correct as circumstances change," and "dynamic steering" that "adapts an agent's behavior based on an individual user's feedback and direction."[1] The post gives a concrete example: "a seller might ask Hunter to rescue their at-risk deals before the end of the quarter. Hunter can turn that objective into a measurable goal, build a plan, and begin working toward it."[1] Analysis: This is the genuinely novel piece of the announcement, more so than the other six agents individually. Most commercially available agents today operate within a single session or task; an agent explicitly designed to carry a goal across weeks, adjusting its own plan as new information arrives while a human retains override authority, is a meaningfully harder engineering problem, which is almost certainly why Salesforce chose to pilot it separately rather than ship it alongside the other six on day one. Salesforce's own framing that "more agents across the portfolio will run on it" over time signals this runtime, not any single named agent, is the more durable product bet underneath the announcement. Published: September 11, 2026.
The Customer Numbers, Read Carefully
Claim: Salesforce backed the launch with specific, named customer performance figures rather than aggregate or anonymised statistics. Source: The post states "50% of Engine's chat inquiries are fully resolved by its help agent, Eva," "60% of Perk's sales pipeline is built by its outbound sales agent, Hunter," "70% of Autism Queensland's administrative requests are resolved by its employee service agent, Paige," "90% of core shopper journeys are handled by Hibbett AI, which went live in six weeks," "4x the conversation volume is now driven by Asana's website agent, Piper," with "customers deploy Piper in 45 days on average" and "79% of Anthropic's conversations that Fin sees are resolved autonomously."[1] Analysis: Naming specific customers, including Anthropic, a direct competitor to Salesforce's own AI ambitions in the sense that it builds foundation models, rather than anonymising results as "a Fortune 500 retailer" is a stronger evidentiary standard than most vendor case studies offer, since each figure is attributable and could in principle be checked against that company's own public statements. It is still, however, a self-selected set of success stories rather than an independently audited or representative sample of Agentforce deployments generally, and the six named companies achieving standout results says nothing definitive about typical or median outcomes across Salesforce's full customer base. Published: September 11, 2026.
Why Job Titles, Not Feature Lists
Claim: Salesforce deliberately built customisability into the product so that each named agent becomes a company's own branded employee rather than a visibly third-party tool. Source: The post states "customers can tailor each agent to their business, including giving it their own name, so the agent becomes an extension of their brand and workforce," and separately notes that every agent "operates within the customer's business rules, permissions, and security," with Agent Script, described as "Salesforce's open-source language for agent behavior," letting "customers combine AI reasoning with deterministic rules for granular control over how agents make decisions and take action."[1] Analysis: Giving an agent a human first name and a renaming option is a small detail with a real commercial purpose: a named, personified agent is easier for a non-technical buyer to place inside a budget conversation or an org chart than an abstract capability, and letting the customer rename it removes the visible Salesforce branding from the actual point of use, which matters for companies that want the agent to read as part of their own team rather than an obviously outsourced tool. This is a marketing and adoption strategy layered on top of the underlying technology, not a technical feature in itself, and it is worth recognising as such when evaluating what is genuinely new here versus what is presentation. Published: September 11, 2026.
How to Actually Try It
Salesforce's own post links directly to each agent's dedicated product page: Casey, Hunter, Marshall, Piper and Fin, with the full portfolio and platform detailed on the Agentforce homepage.[1] Companies wanting employees to work alongside these agents rather than deploy them customer-facing can explore Agentforce Coworker, which the same post confirms now includes Multi-Agent Orchestration, generally available today, alongside AI Skills and Agent Optimizer, both reaching general availability in October 2026.[1]
Common Questions
Sources
- Salesforce, "Salesforce Expands Agentforce With a New Portfolio of AI Agents Built for High-Value Work," official Salesforce Newsroom, September 11, 2026 (updated September 15, 2026). salesforce.com
- Let's Data Science, "Salesforce Launches Job-Ready AI Agents for Sales and Service," September 12, 2026. letsdatascience.com
Read More on Depth Grid
- OpenAI Will Now Rent You the Same Agent Infrastructure That Runs Codex
- Anthropic Just Competed With Its Own Biggest Customers
- Deflection Isn't Resolution: Customer Service AI ROI
- Enterprise AI Agents: Where the Big Checks Go
- Meta's New AI Agent Wants to Book Your Travel and Bills
Article by Mahesh | Depth Grid