A sales coach agent makes the rep better at the conversation; a sales assistant agent makes sure the rep is in the right conversations. The split comes from Force Management's AI ebook, which separates proficiency agents from productivity agents. It is a practical way to sort sales AI tools, because the two kinds need different inputs, do different jobs and should be measured differently.
This entry lays out both, shows where a warm-introduction agent fits, and gives a three-question checklist for evaluating either kind before you buy.
The two kinds, side by side
| Sales coach (proficiency agent) | Sales assistant (productivity agent) | |
|---|---|---|
| Job | Raise the quality of what the rep knows and says | Raise the quantity and precision of the rep's selling time |
| Main inputs | Product content, messaging, competitive intelligence, call recordings, winning deal patterns | CRM records, account and contact data, signals, calendars, relationship data |
| Example tasks | Role-play a discovery call; answer a product question mid-deal; review a call and suggest a better next question; prep talk tracks for a new persona | Build and prioritise an outreach list; personalise messaging; flag a deal that went single-threaded; find the warm path to a buyer and who should ask |
| How to measure it | Ramp time, certification scores, win rate on qualified deals, deal size, call quality | Selling hours recovered, meetings booked, stakeholders engaged per deal, reply and conversion rates, pipeline created |
| Failure mode | Generic advice the rep ignores | More activity that buyers ignore |
Force Management's line that AI augments reps rather than replacing them applies to both columns. A coach that talks over the rep and an assistant that sends as the rep both fail for the same reason: they remove the human judgement the buyer is actually responding to.
Why the distinction matters
A common risk is buying sales AI one tool at a time and then wondering why productivity has not moved. Gartner found that AI saves sellers about 4.8 hours a week, yet 72% of sales organizations report low reinvestment of that time, and the teams that do reinvest are 2.2x more likely to exceed customer growth goals (Gartner, 19 May 2026). Coaches and assistants attack that gap from opposite ends. A coach improves what happens inside the hours; an assistant decides where the hours go. Buy only one kind and half the problem stays open. See the AI time reinvestment gap.
The distinction also stops the most common procurement mistake: judging an assistant on coaching metrics, or a coach on activity metrics. A coach will never book more meetings on its own, and an assistant will never make a weak rep strong in the room.
Where a warm-introduction agent fits
A warm-introduction agent is a sales assistant. Force Management describes productivity agents as helping reps find "who to talk to and when", and that is precisely the job. Boomerang's agent, Rudy, answers it from relationship data rather than contact data:
- Who. Which of your executives, employees, investors, advisors, board members, customer champions and partners actually knows the buyer, graded from more than 80 relationship signals.
- Who should ask. The internal hop: the person inside your company with the strongest tie to that connector, so the request comes from whoever is most likely to get a yes.
- When. Rudy reads the CRM, calendars, call recordings and Slack, so it knows when a deal moves, goes single-threaded or loses a champion, and raises the path at that moment.
What keeps it an assistant rather than an autonomous seller is the consent model. Rudy proposes. The rep approves the plan. The person who owns the relationship approves the ask and sends it from their own account. Rudy never sends on anyone's behalf. For the full design, see AI agent for warm introductions.
How to evaluate either kind of agent
Force Management suggests pressure-testing any AI investment against a small set of questions. Paraphrased, they come down to three:
- What problem does it solve, specifically? Name the gap: slow ramp, weak discovery, too few meetings, single-threaded deals. If a vendor cannot tie the agent to one, it is a feature looking for a buyer.
- What are its inputs and guiding principles? What data does it read, what does it refuse to do, and who approves its output? A coach trained on generic content gives generic advice. An assistant with no relationship data can only recommend cold outreach.
- Which KPIs will prove it worked? Pick them before the pilot, from the right column of the table above, and set a baseline. If the only metric is usage, you are measuring adoption, not value.
Worked example
A mid-market team rolls out a coaching agent for discovery calls and a warm-intro assistant in the same quarter. The coaching agent is judged on how often reps reach the economic buyer in discovery and on stage-two conversion. The assistant is judged on how many open deals gain a second stakeholder through a warm path, and how many of those introductions turn into meetings. Each tool gets a metric it can actually move, and the review at quarter end is about outcomes, not logins.
Bottom line
Sort every sales AI tool into coach or assistant before you evaluate it. Coaches should be judged on what reps know and say; assistants on where reps spend their time and who they reach. A warm-introduction agent belongs in the assistant column, and its test is simple: does it put the rep in front of the right person sooner, while leaving people in charge of every ask?
Frequently asked questions
What is the difference between a sales coach and a sales assistant AI agent?
In Force Management's framing, a sales coach is a proficiency agent: it makes reps smarter about product, market, content and what to say. A sales assistant is a productivity agent: it builds lists, personalises outreach and helps reps find who to talk to and when. One improves the conversation; the other gets the rep into the right ones.
Is a warm introduction agent a sales coach or a sales assistant?
A sales assistant. It does not teach the rep what to say; it finds who in the company knows the buyer, who should make the ask and when. Boomerang's Rudy does this from relationship data, proposes the path, and leaves approval and sending to the people who own the relationship.
How should I measure a sales coaching AI agent?
On proficiency outcomes: ramp time, certification or assessment scores, call quality, win rate on qualified deals and deal size. Set a baseline before the pilot. Usage and logins show adoption, not value, and a coach should not be judged on meetings booked, which it cannot move directly.
How should I measure a sales assistant AI agent?
On productivity outcomes: selling hours recovered, meetings booked, stakeholders engaged per deal, reply and conversion rates, and pipeline created. For a warm-introduction assistant, track how many deals gain a new stakeholder through a warm path and how many of those introductions become meetings.
Do I need both kinds of AI agent?
Usually, yes; relying on one kind is a common risk. Gartner found AI saves sellers about 4.8 hours a week, yet 72% of sales organizations report low reinvestment of that time. A coach improves how those hours are used in conversations; an assistant decides where they go. Buying only one leaves half of the gap open.
What questions should I ask before buying a sales AI agent?
Three, paraphrasing Force Management. What specific problem does it solve? What are its inputs and guiding principles, including what it refuses to do and who approves its output? And which KPIs will prove it worked, set with a baseline before the pilot starts?
