Salesforce Ships Hunter and Piper: What Named AI Sales Agents Mean for Outbound
On September 11, 2026 Salesforce launched seven named Agentforce agents, including Hunter for outbound pipeline and Piper for inbound qualification. The naming is not marketing fluff: it marks the shift from AI tasks to AI job roles. What actually shipped, what the pilot claims are worth, and what a Hunter would do to German-language outreach.
Tools in this post
Key takeaways
- Salesforce launched seven named AI agents on September 11, 2026, each scoped to a job rather than a task. For sales: Hunter builds outbound pipeline from research to outreach (pilot, GA November 2026) and Piper qualifies inbound interest (generally available now).
- The naming signals the real shift: agents are no longer features inside software but coworkers with a role, a scope and a supervisor. That framing, not the technology, is what will change how sales teams are structured.
- The headline pilot claim, 60 percent of one customer's pipeline built by Hunter, is a vendor-selected best case. The credible core is narrower: research, list work and first drafts at scale, with humans on message strategy and conversations.
- For DACH the checkpoint logic is unchanged and now urgent: an outbound agent without native German review scales machine tone into a market that punishes it. The preparation, clean data, sharp ICP, documented process, is identical whether your Hunter comes from Salesforce or your own stack.
Two weeks after Claudeforce made Claude the reasoning layer of the Salesforce world, Salesforce shipped the other half of the story. On September 11, 2026, just ahead of Dreamforce, it launched seven named Agentforce agents, each built for a specific business function. Two of them do sales. Hunter builds outbound pipeline, from account research to outreach, currently in pilot with general availability planned for November 2026. Piper turns inbound interest into qualified opportunities and is generally available now. The others, Casey, Carter, Paige, Marshall and Fin, cover service, commerce, internal helpdesk, supply chain and customer experience.
It is easy to file this under product marketing. That would miss what actually changed.
Why the names matter
Software features do not get first names. Coworkers do. By shipping agents as named roles rather than capabilities, Salesforce is packaging a claim about how work is organized: an agent has a job, a scope, and a human it reports to, the same way a new hire does. Alongside the agents, Salesforce previewed a six-pillar governance framework, the Trusted Enterprise AI Harness, covering context, agency, action, governance, security and models, with broad availability expected in early 2027. Read together, the message to enterprises is: you can now hire, scope and supervise digital workers the way you manage people.
That framing will spread far beyond Salesforce, because it answers the question that has stalled agent adoption in every mid-size company we talk to: not "can the model do it" but "who is responsible when it acts". A named agent with a defined scope and a supervisor is an answer organizations understand.
What Hunter actually does, and what the claims are worth
Hunter is described as handling outbound pipeline development end to end: researching accounts, building target lists, drafting and sending outreach, collaborating with sellers over multi-week cycles rather than executing single tasks. One pilot customer reports that 60 percent of their sales pipeline is now built by Hunter.
Treat that number the way you would treat any vendor-selected pilot statistic: as the best case of the best customer, with "built by" doing a lot of undefined work. Is a meeting "built by Hunter" when the agent researched the account and drafted the first email that a human edited and a human closed? By that definition, well-run teams have had 60 percent of their pipeline "built by" their systems for years. The credible core of Hunter is narrower and still significant: the research, list building and first-draft layer of outbound, the exact jobs we described as dependable in our piece on AI agents in sales, now packaged as a product inside the CRM instead of assembled from tools.
Piper is the less flashy and more immediately real of the two: inbound follow-up and qualification is high-volume, low-ambiguity work where minutes decide conversion, which is exactly the profile where agents already outperform delayed humans.
The DACH reading
Three things are worth saying clearly for the German-speaking market.
First, the checkpoint logic does not move. An outbound agent that researches, drafts and sends without native review will scale machine-tone German into a market that detects it in one sentence and punishes the sender's domain for it. Whatever Hunter can do in English at volume, in DACH the final-review gate and the legal layer, channel rules per country, opt-out handling, stay human and rule-based. Salesforce's own governance framing, agents acting within granted scopes, points the same direction.
Second, most of the DACH mid-market will meet this pattern outside Salesforce. Named-role agents will arrive in every stack, assembled from the same parts: an agent runtime, MCP connections into CRM, enrichment and sequencing, and a scope definition. The architecture is the one we mapped in the future of GTM engineering, and the preparation is identical whether you buy your Hunter from a vendor or build one: clean data, a sharp ICP, a documented process an agent can be scoped to.
Third, the adoption numbers say this is no longer early. Salesforce's own State of Sales research, published alongside the launch, has 54 percent of sellers already using agents and nearly nine in ten planning to by 2027; we pulled the numbers apart, including the ones that cut against the hype, in our State of Sales analysis. The window in which running a disciplined, agent-assisted outbound system is a differentiator rather than table stakes is measured in quarters now.
What to do with this
Nothing about this launch requires buying anything. It requires deciding, sooner than planned, how your own outbound motion answers the questions Hunter's packaging answers: what is the agent's scope, where are the human gates, who supervises, and what data does it stand on. Teams with those answers will plug named agents in, from whichever vendor, and compound. Teams without them will hand a fast agent a fuzzy ICP and a stale CRM, and scale the mess. The unglamorous prerequisites, the ones from our agency question checklist and DACH benchmarks, just became urgent.
Frequently asked questions
What are Salesforce's named Agentforce agents?
Seven AI agents launched on September 11, 2026, each scoped to a business function: Hunter for outbound sales pipeline (pilot, GA planned November 2026), Piper for inbound qualification, Casey for customer service, Carter for e-commerce, Paige for internal IT and HR, Marshall for supply chain, and Fin for customer experience. Most are generally available; Hunter remains in pilot.
What does the Hunter agent do?
Hunter handles outbound pipeline development inside Salesforce: account research, target list building, outreach drafting and multi-week collaboration with sellers. A pilot customer reports 60 percent of pipeline built by Hunter, a vendor-selected figure worth reading skeptically. The dependable core matches what agents do well generally: research, lists and first drafts, with humans on strategy and conversations.
Do named AI agents replace SDRs?
They replace the execution layer of the SDR job, research, list work, drafting and follow-up timing, not the judgment layer. Message strategy, final review of outgoing communication and live conversations stay human, and in the German-speaking market the native-language review gate is non-negotiable because recipients detect and punish machine tone quickly.
Is Agentforce relevant for companies not using Salesforce?
As a product, only if you are in that ecosystem. As a pattern, yes: named agents with a defined job, scope and human supervisor will arrive in every stack via agent runtimes and MCP connections to CRM, enrichment and sequencing tools. The preparation is vendor-independent: clean data, a sharp ICP and a documented, scopeable process.