---
title: "Coach reps on a new product with agents"
slug: "coach-reps-on-a-new-product-with-agents"
updated: 2026-11-11T15:31:52Z
published: 2026-11-11T15:31:52Z
canonical: "help.gong.io/coach-reps-on-a-new-product-with-agents"
---

> ## Documentation Index
> Fetch the complete documentation index at: https://help.gong.io/llms.txt
> Use this file to discover all available pages before exploring further.

# Coach reps on a new product with agents

> **Who can use this**: Any team member
> 
> **Available on**: Gong Foundation
> 
> **Ideal for**: Sales manager
> 
> **Required permissions:** Manage company agents, skills, and automations

After a launch, every rep has to explain the new feature in their own words. Managers cannot listen to every call, so they coach from guesswork, and the messaging slowly drifts away from what you approved.

You can have an agent do the first pass. It compares how your reps talked about the feature with your approved enablement material and talk track, and gives the manager a short note: What is working and what is missing, with quotes and specific fixes, but no pass or fail.

You can set up an agent to coach reps on a new product in three ways. What differs is who starts the agent and where the note goes:

| Way to build | What you get | Choose it when |
| --- | --- | --- |
| Start with the Assistant | A manually triggered agent, drafted for you from a description in plain words. You start it yourself and read the coaching note. | You want a working agent quickly, and you will decide when to run it. |
| Build an automatically triggered agent | An agent that runs by itself each time your tracker detects a rep explaining the new product on a call, and sends the coaching note to the person you choose. | You want coaching on every relevant call, without having to pick the calls yourself. |
| Build a manually triggered agent | A manually triggered agent that you build from scratch, with your own instructions and enablement files. You start it yourself and read the coaching note. | You want to start from scratch, or to test your enablement files before you automate anything. |

## Start with the Assistant

Describe the agent in plain words, and the Assistant drafts it for you. You check the draft and save it.

The Assistant builds a manually triggered agent. You start it yourself, and the coaching note appears for you to review and share. A manually triggered agent has no trigger and no notify step, so the Assistant cannot send the note to anyone for you. You can [run it manually directly from the Assistant](/v1/docs/run-agents-in-the-gong-assistant). To have an agent run on its own, see [Build an automatically triggered agent](/docs/coach-reps-on-a-new-product-with-agents#build-an-automatically-triggered-agent).

**To build the agent with the Assistant:**

1. From the left sidebar, click **Agents**.
2. Click **New agent**, and select **Start with the Assistant**.
3. In the **Build an Agent** chat, paste the following prompt. Replace `[FEATURE NAME]` with the name of your new product or feature, and send it:

```plaintext
Create an on-demand agent called "New product rep coaching" that I run when I want to check how our reps explained [FEATURE NAME] on customer calls.

It should compare what the reps said with the approved enablement files I attach to the agent. For each call, write a short note for the rep's manager:
1. Feature coverage: covered, partial, or not meaningfully covered.
2. What went well: two or three points, each with a short quote from the call.
3. Gaps: each with the rep's quote and what the approved material says instead.
4. Next-call practice: one to three specific fixes.
5. Anything that needs verification.

If the feature was not meaningfully explained on a call, say so and stop for that call. Treat the attached files as the source of truth, and say "needs verification" instead of filling gaps from general knowledge. Do not give a pass or fail, and do not judge a rep's skill or performance. Max 300 words per call. Tone: constructive and specific, written for the manager.
```

1. Review the agent that opens next to the chat. Check that the description and instructions match what you asked for.
2. When the Assistant asks to save the agent, reply `yes`, and click **Save** in the dialog.
3. When the Assistant says the agent has been created, click the link to open it.
4. (Optional) To change what the agent does, edit the **Description** or **Instructions**.
5. Click **Add files**, and upload your approved enablement material, for example the feature brief, the talk track, and a short coaching checklist. In the **Upload files** window, select files from your computer or Google Drive. You can add up to ten files of 20 MB each, as PDF, DOCX, TXT, MD, RTF, or CSV. Add only current, approved files. When the files finish scanning, click **Add files**. Then, in **Instructions**, describe what each file contains and how the agent should use it.
6. (Optional) To give the agent reusable guidance, click **Add skills**.
7. (Optional) To let everyone in your company use the agent, in the top right, click **Personal**, and select **Company**.
8. Click **Save**.

The prompt tells the Assistant:

- The name of the agent, and that you start it yourself.
- What to compare: how your reps explained your feature, against your approved files.
- What to write: coverage, what went well, gaps with quotes, and next-call practice.
- What to do when the evidence is weak: say so, and mark anything the files do not settle as needing verification.
- What to avoid: a pass or fail rating, and any judgment of a rep's skill or performance.

> [!NOTE]
> Note:
> 
> The Assistant rewrites your prompt into its own instructions. For example, it may rename the sections or add rules of its own. To change anything, click **Keep editing** in the **Save agent?** dialog, and tell the Assistant what to change.

## Build an automatically triggered agent

This agent runs by itself. Each time your tracker detects a rep explaining the new product on a customer call, the agent compares the call with your enablement files, and sends a coaching note to the person you choose. For example, a coaching note sent as a Slack message looks like this:

> **Example coaching note**
> 
> New product coaching signal: Harbor Analytics
> 
> - **Feature coverage:** Partial. The rep explained what the feature does but not who it is for.
> - **What went well:** The rep tied the feature to the customer's reporting problem, with a clear example.
> - **Gaps:** The rep said the feature is "included for everyone." The talk track says it is available on selected plans only.
> - **Next-call practice:** Confirm the customer's plan before describing availability.
> - **Needs verification:** Whether the customer's plan includes the feature.
> 
> AI-generated coaching aid. Check it against the call before you act.

You need a published AI tracker that detects your rep explaining the new product. It should match the rep's explanation, not a customer asking about the feature. To create one, see [Create AI trackers](/v1/docs/create-ai-trackers).

> [!NOTE]
> Note:
> 
> The note is a coaching aid, not a rating. It does not give a pass or fail, so check its quotes against the call before you share feedback with the rep. Automatically triggered agents consume Gong credits, so keep the tracker narrow.

**To build the agent:**

1. From the left sidebar, click **Agents**.
2. Click **New agent**, and select **Start from scratch**. Then select **Automatically triggered**.
3. Click the agent name at the top, and enter a name, for example `New product rep coaching`.
4. In the **Add a trigger** panel, select **AI tracker**. Then click **Select AI tracker**, and select your tracker.
5. Click the **Select step** card, and select **Generate with AI**.
6. In the **Instructions** field, paste the following instructions. Replace each `[FEATURE NAME]` with the name of your new product or feature.

```plaintext
Role: You are a product-enablement sales coach reviewing the triggering customer call after our rep discussed [FEATURE NAME].

Use: the triggering call, limited available CRM context, and the attached approved enablement, product-reference, talk-track, and coaching-rubric files.

Source of truth: use the attached approved files to evaluate the product explanation. Do not assume access to other documents, systems, product usage, training platforms, HR data, or manager hierarchies. If the files do not establish an answer, mark it "needs verification."

Produce a concise coaching alert for a manager or enablement reviewer:
1. Feature coverage: covered / partial / not meaningfully covered.
2. What went well: up to three bullets, each with a short quote or precise call moment.
3. Gaps or risks: each with the rep's quote or moment and the approved expectation it should be compared with.
4. Next-call practice: one to three specific behaviors or revised phrases.
5. Verification needed: any product, packaging, availability, or enablement claim that requires confirmation.

Rules:
- Do not send a pass/fail judgment.
- Do not infer competence, intent, HR status, compensation impact, or promotion readiness.
- Do not invent product claims, customer needs, or next steps.
- If [FEATURE NAME] was not meaningfully discussed, say so and do not create a coaching alert beyond that fact.
- Keep the output under 300 words.
- Write constructively and specifically for the manager or enablement reviewer.
```

1. In the same step, under **Files**, click **Add files**, and add your approved enablement material, for example the feature brief, the talk track, and a short coaching checklist. Add only current, approved files.
2. Click the **Add step** icon (+) below the **Generate with AI** step. Then choose how you want to notify people:
  - To notify by Slack message, select **Send a Slack message**:
    1. To message a person, click **Select recipients**, and select **CRM role**, **Manager of CRM Role**, or **Team member**. Then select who should get the note, for example the rep's manager or your enablement lead.
    2. (Optional) To message a channel, click **Select channels**, and select one of your company's channels, for example a shared enablement channel.
    3. In **Message**, enter `New product coaching signal:`. Click **Add variables**, and select the account name from the trigger data.
    4. On a new line, click **Add AI response**.
    5. On another new line, enter `AI-generated coaching aid. Check it against the call before you act.`
  - To notify by email, select **Send an internal email**:
    1. Click **Select recipients**, and select **CRM role**, **Manager of CRM Role**, **Team member**, or a mix. Then select who should get the note.
    2. In **Subject**, enter `New product coaching signal`.
    3. In **Message**, write a short introduction, for example `A rep explained the new product on a recent call. Here is a coaching note.`
    4. On a new line, click **Add AI response**.
3. Click **Save**.

To turn the agent on and check that it works, see [Create an automatically triggered agent](/v1/docs/create-an-automated-agent).

## Build a manually triggered agent

Build the agent yourself when you want to write your own instructions from scratch and add your enablement files. Like the agent the Assistant builds, you start it yourself, and it has no trigger and no notify step. It is also a good way to test your enablement files on a few calls before you automate anything.

**To build the agent:**

1. From the left sidebar, click **Agents**.
2. Click **New agent**, and select **Start from scratch**. Then select **Manually triggered**.
3. In **Description**, enter what the agent does, for example `Compares how reps explained a new product with approved enablement files, and returns strengths, gaps, and next-call practice.`
4. In **Instructions**, paste the following instructions. Replace each `[FEATURE NAME]` with the name of your new product or feature.

```plaintext
Role: You are a product-enablement sales coach assessing how sales reps explained [FEATURE NAME] in customer calls.

Use: the selected call or calls, plus the attached approved enablement, product-reference, talk-track, and coaching-rubric files.

Source of truth: treat the attached approved files as the source of truth for what [FEATURE NAME] does, who it is for, the approved value message, required talk-track elements, limitations, and availability. If the calls or files do not establish something, say "not established" or "needs verification." Do not fill gaps from general knowledge.

Task: If you are given more than one call, assess each call separately. For each call:
1. Decide whether the rep gave a substantive explanation of [FEATURE NAME]. If not, say that the feature was not meaningfully covered and stop the assessment for that call.
2. Compare what the rep said with the approved files.
3. Identify what was clear, accurate, and effective.
4. Identify missing, unclear, unsupported, or potentially inaccurate elements.
5. Give one to three specific fixes or practice suggestions for the rep's next call.

Evidence rules:
- Support every strength or gap with a short quote or precise moment from the call.
- For every gap, name the relevant expectation from the attached source file.
- Do not infer rep intent, customer sentiment, skill level, performance rating, compensation impact, or HR status.
- Do not invent product capabilities, customer requirements, or next steps.
- Do not use a pass/fail judgment.
- If a call does not contain enough evidence, say so plainly.

Output:
- Feature coverage: covered / partial / not meaningfully covered, with one-sentence reason.
- What went well: two or three bullets with call evidence.
- Gaps or risks: each bullet must include the call evidence and the approved expectation it should be compared with.
- Next-call practice: one to three concrete actions or revised behaviors.
- Verification needed: list anything that requires Product or enablement confirmation.

Keep the output under 300 words per call. Write for the manager or enablement reviewer in a constructive, specific tone.
```

1. Click **Add files**, and upload your approved enablement material, for example the feature brief, the talk track, and a short coaching checklist. Add only current, approved files. Then, in **Instructions**, describe what each file contains and how the agent should use it.
2. (Optional) To give the agent reusable guidance, click **Add skills**.
3. (Optional) To let everyone in your company use the agent, in the top right, click **Personal**, and select **Company**.
4. Click **Save**.

To run the agent, see [Run agents in the Gong Assistant](/v1/docs/run-agents-in-the-gong-assistant).

## More agents you can build

You can build more agents that send coaching notes on other parts of a call. Each one below says what the agent does and when it runs, and gives the instructions to paste. To build one, follow the steps in [Build an automatically triggered agent](/docs/coach-reps-on-a-new-product-with-agents#build-an-automatically-triggered-agent), select the trigger named for that agent, and paste its instructions into the **Instructions** field. An agent that starts from an AI tracker needs a published tracker that detects the moment it describes.

****Overselling alert agent****

Catch promises a rep may not be able to keep. When a rep appears to commit to a capability, a timeline, or an outcome on a call, the agent alerts product or enablement leadership with what was promised, the rep's exact words, and why it reads as a firm commitment. The agent does not judge whether the promise is accurate. It leaves that to the person who reads the alert.

- Trigger: An AI tracker that detects a rep promising a capability
- Notify: Product or enablement leadership, by Slack message or email

```plaintext
Role: You are a compliance assistant flagging over-commitment on calls.

Use: the call that triggered this, plus limited CRM context. Do not assume access to fields, boards, or systems beyond that.

Produce: a short alert when a rep appears to promise a capability, timeline, or outcome.

Focus on: what was promised, the exact phrasing, and why it reads as a firm commitment (its specificity or certainty). If the rep only described existing capability and made no forward promise, say so and do not raise an alert.

Format: three bullets: the promise, the quote, why it reads as a firm commitment. Under 40 words.

Tone: plain and factual, written for leadership.

Don't: invent information not in the call or email; assume what is or is not shippable; judge whether the promise is accurate. Leave that judgment to the human who reads the alert.
```

****Objection-handling coaching agent****

Coach reps on the objections they actually hear. When a customer raises an objection on a call, the agent tells the rep's manager what the objection was, how the rep responded, with a quote, whether the response was effective or weak, and a better approach if it was weak. If the customer only voiced a passing concern, the agent says so instead of rating it.

- Trigger: An AI tracker that detects an objection
- Notify: The rep's manager, by Slack message or email

```plaintext
Role: You are a sales coach evaluating how a rep handled an objection on a call.

Use: the call that triggered this, plus limited CRM context. Do not assume access to fields, boards, or systems beyond that.

Produce: a short read of the objection and how well the rep handled it.

Focus on: the objection raised, how the rep responded, and whether the response was effective, with a better approach if it was weak. If no real objection was raised on this call and the customer only voiced a passing concern, say so and do not force a rating.

Effectiveness rubric: "Effective" means the rep addressed the specific concern and the customer moved on or agreed. "Weak" means the rep changed the subject, gave a generic answer, or the customer repeated the concern.

Format: four bullets: objection, rep's response with one short quote, quality read (effective/weak, per rubric), suggested improvement. Under 70 words.

Tone: constructive and specific, written for the manager.

Don't: invent information not in the call or email; judge a response the call does not contain; invent a better line the rep could not have known.
```

****Coaching-topic agent****

Follow how reps handle the topic you are coaching on, such as a value message or a pricing conversation. When the topic comes up on a call, the agent sends the rep's manager a short quote showing how the rep handled it, what went well, what was missing, and what to coach next. If the topic did not really come up, the agent says so instead of forcing a coaching point.

- Trigger: An AI tracker that detects the coaching topic
- Notify: The rep's manager, by Slack message or email

```plaintext
Role: You are a sales coach assessing how a rep handled a specific topic on a call.

Use: the call that triggered this, plus limited CRM context. Do not assume access to fields, boards, or systems beyond that.

Produce: a short read of how the rep handled the topic, with one coaching suggestion.

Focus on: how the rep addressed the topic (for example a value message, a pricing conversation, a discovery area), what went well, and what to coach. If the topic did not really come up on this call, say so plainly and do not force a coaching point.

Format: one short quote showing how the rep handled the topic, then three or four bullets: what happened, strength, gap, what to coach next. Under 70 words.

Tone: constructive and specific, written for the manager.

Don't: invent information not in the call or email; assess a topic the call did not cover; give generic coaching that would fit any call.
```

****Stage-exit feedback agent****

Give reps feedback while a stage is still fresh. When an opportunity leaves a stage in your CRM, the agent reviews the recent calls on the account and sends the rep a short note: One moment from the calls, what was strong, what was thin, and the one thing to fix, for example whether discovery went deep enough. If the calls do not show enough to judge, the agent says so. To set up a CRM change trigger, see [Create an automatically triggered agent](/v1/docs/create-an-automated-agent).

- Trigger: A CRM change, when the opportunity leaves a stage
- Notify: The rep, by Slack message

```plaintext
Role: You are a sales coach giving fast feedback as the account moves out of a stage.

Use: the recent calls on the account, plus limited CRM context. Do not assume access to fields, boards, or systems beyond that.

Produce: immediate, specific feedback tied to the stage just completed.

Focus on: whether the work that stage requires was actually done (for example discovery depth before leaving discovery), and the one thing to fix. If the recent calls do not show enough to judge the stage work, say so instead of guessing.

Format: one moment from the calls as evidence, then two or three bullets: what was strong, what was thin, the one fix. Under 50 words.

Tone: constructive and direct, written for the rep.

Don't: invent information not in the calls; assume stage requirements the team has not set; invent gaps.
```

For agents for other teams, with their instructions, see the [Custom Agents prompt library](https://academy.gong.io/learn/course/custom-agents-prompt-library/custom-agents/prompt-library) in Gong Academy.

Any team member can access this feature if they have the required permissions.

Gong’s core platform for gaining advanced visibility into your revenue operations. Also the prerequisite license needed for Gong’s add-on packages. Access to features depends on your [company’s plan and your assigned seat](/v1/docs/plans-and-seats).

A team member who manages a team of sales reps.