Why we built it

• Deliver consistent, personalized feedback without adding manager workload.

• Systematically track and coach skills like objection handling and stakeholder discovery.

• Spot trends and coaching opportunities across teams with centralized insights.

Overview

Problem

Multiple teams have diverse enablement needs. Managers are spread thin, struggling to provide consistent one-on-one coaching and feedback. There's a backlog of enablement tasks and a need for more efficient, personalized training.

Solution

An AI-powered sales and success coaching bot that analyzes Gong call transcripts. The bot evaluates key aspects like discovery questions, objection handling, stakeholder identification, and agenda setting. It generates tailored feedback for different teams and allows managers to send automated, personalized coaching messages to reps.

How it works

When: A new call uploaded is to Gong.

Then:

1. Send call transcript to AI for analysis and populate feedback tables for managers.

2. A webhook will call the needed API endpoint to extract the call transcript. In this case, the URL of the endpoint is: https://us-9905.api.gong.io/v2/calls/transcript (A Google search or tools like Postman or RapidAPI can help you find the right endpoint you need).

3. New call transcript summaries are added to a central sheet.

4. The Zap then leverages Paths to branch into different scenarios. In this example, the different paths are based on the different roles on the team as they have different needs and areas of focus during sales calls:

5. We provide different coaching based on teams

• The priority for Inside Sales is coaching them on problem identification.

• The priority for Account Managers and Customer Success is coaching them on stakeholder identification.

💡 Tips

Finding API endpoints for webhooks: A Google search or tools like Postman or RapidAPI can help you find the right endpoint you need.

Using Formatter: Formatter is your friend. Often the output of a step isn’t in the format we need it to be. The formatter step helps you tackle that challenge. You’ll notice this Zap has a couple of formatter steps. Speaking of which… The formatter step right before the AI step truncates the transcript. This is because large input messages into AI can sometimes time out the ChatGPT request, consume a lot of tokens, and/or affect the quality of the summary produced by ChatGPT.

Crafting your AI prompts for assistant instructions: Be very detailed in the AI prompt with clear examples of what type of things to look for in the call and what type of feedback to provide. The Zap uses several AI steps that provide different outputs. See below for each one and the prompts associated with them:

• Transcript summary: This is the initial AI step that summarizes the call transcript. (View example)

• Manager summary: This provides a summary to the respective managers based on the area of focus. (View example)

• Rep message: This is a draft of the message to be sent to the rep by the manager that highlights strengths and areas of opportunity. (View example)