Salesforce Built 7 AI Agents. None Coach Your Closer.

On September 11, 2026, Salesforce announced a portfolio of seven “job-ready” AI agents built for specific business functions. Two are for sales: Hunter, an outbound agent that works pipeline from research through outreach over weeks and months, and Piper, an inbound agent that engages and qualifies leads across websites and inboxes. The other five cover customer service, IT, commerce, supply chain, and customer experience.

Every major tech outlet covered what those seven agents do. Nobody asked what the entire lineup leaves out.

What the Launch Actually Covers

Salesforce is not being modest about scale. According to the September 11 press release, Agentforce has delivered 7 billion Agentic Work Units (AWUs) across Agentforce and Slack, including 3.2 billion in Q2 2026 alone. The customer results are concrete: 60% of Perk’s sales pipeline is now built by Hunter, the outbound agent.

That is a real outcome. If you run a Salesforce shop, or you are evaluating whether to expand your Agentforce footprint, Hunter and Piper solve problems that sales managers have complained about for years. Automating outbound research and follow-up is real leverage. Qualifying inbound leads around the clock without a human on standby is real leverage. The pipeline automation use case is legitimate and the scale numbers suggest that adoption is already well past the early-adopter phase.

This article is not a takedown of those tools. It is about what comes next, once those tools start working.

What the Launch Does Not Cover

Hunter gets a lead to the calendar. Piper qualifies an inbound inquiry and passes it to the team. When both hand off to a closer, the agent’s job is done. The call is on the rep.

Not one of the seven Agentforce agents touches what happens inside the conversation. There is no agent reviewing what your rep said in the first two minutes. No agent surfacing which reps crumbled on the same objection three calls in a row. No agent telling you that your second-best closer is losing the frame somewhere around minute eight on every discovery call. Salesforce built AI for everything leading up to the conversation and for everything downstream of it. The conversation itself is still a black box.

That is not a gap in Agentforce specifically. It is a gap in the entire AI sales stack. Five years of investment and the most sophisticated pipeline automation in the industry’s history, and the thing that actually determines whether the deal closes is still sitting entirely with the rep, unobserved and uncoached.

The better your pipeline machine works, the more expensive that gap becomes.

The Take

The AI sales industry has spent the better part of a decade solving the lead problem. It has largely ignored the close problem.

That is not a neutral observation. It has a specific consequence: a sales team that adds Hunter and Piper is going to generate more qualified meetings. If those meetings hit a rep population with the same coaching process you have today, the close rate stays where it is. You do not get a better outcome. You get a bigger pipeline running through the same leak.

The data on this is not ambiguous. MySalesCoach’s State of Sales Coaching 2026, drawn from more than 1,000 respondents, found that teams receiving weekly coaching hit quota at 76%. Teams coached quarterly or less hit it at 47%. That is a 29-point spread, and the only variable is how often managers are working with their reps on the craft of selling.

The same report found that 66% of sales managers have never received any training on how to be an effective coach. So the reps are undercoached because the managers were never taught to coach, and the industry’s answer is to ship more pipeline automation. More meetings into a broken coaching loop does not close the performance gap. It gives that gap more room to do damage.

This is the move nobody is publicly making: Agentforce proves the market has a working answer to “how do we get more qualified meetings on the calendar.” It has no answer for “how do we make sure the closer is ready when the meeting happens.” Those are two different products, and only one of them shipped last Thursday.

Why This Is a Monday Problem

If you are already seeing Hunter and Piper in a vendor conversation, or if your ops team is evaluating expanding your Agentforce footprint, the pipeline case for those tools is real and you should evaluate it on its merits.

The Monday question is: what does your current coaching infrastructure look like? Not in the org chart. In practice, last month. How many calls per rep did you actually review? How many reps got specific, actionable feedback on what they said inside a conversation? If your answer involves “I try to listen when I can” or “we do call review when we have time on Fridays,” you are about to have more pipeline than your coaching can absorb. This is the coaching gap that most managers understand in theory and underestimate in practice.

Piling more meetings on top of that situation is not a strategy. It is a stress test.

What to Do This Week

The coaching layer does not require a new platform to start building. Here is where to focus between now and next Monday:

1. Count your actual call reviews, not your intended ones. Pull last month’s numbers: how many calls per rep did you listen to, and how many got specific feedback the same week? For most managers, the number is well below what they would estimate. That gap is your coaching debt. Naming it is the first step to managing it.
2. Find one objection pattern across recent lost deals. Pull five lost calls from the last 30 days and look at what the rep said in the final ten minutes. If the same theme appears three or more times, you have a coachable issue that is currently invisible to the team. That is your first coaching target, this week, before you add one more meeting to the pipeline.
3. Set a call review floor for Q4 and hold it. The 29-point quota attainment gap between weekly-coached and quarterly-coached teams is not an argument for perfection. It is an argument for consistency. Decide on a call review cadence you can actually keep, whether that is two calls per rep per week or one, and make it a standing commitment, not a whenever-I-have-time activity.
4. Separate coaching from performance management. Most managers conflate listening to a call with having a difficult conversation about results. They are different activities and they require different mental frames. Call review is a teaching moment. Keep it specific and clinical: what worked, what did not, one adjustment to try next call. If it becomes a broader performance conversation, the coaching value disappears. Keep the two activities in separate meetings.
5. Pay attention to what your reps are not saying. The hardest patterns to catch in a call review are the omissions: the moment the rep accepted a vague timeline without pinning it down, the moment they had an opening to establish urgency and let it pass, the moment they heard a clear objection and redirected instead of addressing it. What a rep does not say is often more diagnostic than what they do. Train yourself to listen for the silence. A related breakdown in how teams approach this is described in our piece on the manager confidence gap.

None of these steps require a platform decision. They require a consistent process and a manager who has enough visibility into conversations to coach on them specifically. The bottleneck is almost never effort. It is always visibility and bandwidth.

What Salesforce Left on the Table

Seven agents across every major business function. Hunter fills the calendar. Piper warms inbound. Casey handles service. Paige covers IT and HR. Carter runs commerce. Marshall manages supply chain. Fin handles customer experience.

The pattern is clear when you list them out. Agentforce now has AI coverage for every phase of the customer journey except one: the live sales call where a human closer is trying to get a yes from a qualified buyer that the pipeline machine just delivered. Salesforce did not miss this by accident. Building an agent that coaches a rep in real time is a genuinely hard problem, and the liability questions around that kind of intervention are real. But the gap is there regardless of the reason for it, and it lands on your desk as a management problem the moment Hunter starts booking meetings at scale.

For managers already thinking through this, our breakdown of why AI sales tools keep missing on ROI covers the structural reasons the coaching layer keeps getting skipped.

eNZeTi is the layer that Agentforce left out. It processes recorded calls and tells a sales manager specifically what happened, what patterns are appearing across reps, and where to spend coaching time, without the manager needing to listen to every recording individually. When Hunter books a meeting and Piper qualifies the lead, someone still needs to figure out why the closer won or lost the call. That is the problem eNZeTi is built to solve, and it is the one the seven-agent lineup announced on September 11 does not touch.