Configuring your sales coach

1. Configuring your coaching assistant

Welcome back! The first version of our Sales Call Coach agent is live. Now, we'll teach it to parse conversations, tag key moments, and deliver coaching feedback that actually helps reps improve.

2. From messy text to structured coaching data

Raw call transcripts are messy. They're dense blocks of text without clear structure, making it more complicated for your AI to identify who said what or when key moments happened. By parsing transcripts into conversation turns and tagging them with metadata like speaker labels, rep questions, customer objections, and buying signals, unstructured text is transformed into analyzable data. By specifying this parsing step in your custom agent, we'll get better results, leading to a more consistent experience of our sales coach agent.

3. Tagging responses for coaching intelligence

What should we teach our AI agent what to look for? Here are some examples: If a sales rep asks three discovery questions in a row, that could be considered a strength. If they miss a buying signal or rush past an objection, that's a coaching opportunity. If the call ends without a clear next step or follow-up action, tag it as a missed closing action. Structured tagging improves the agent's coaching feedback by giving it the context it needs to evaluate tone, objection handling, and closing effectiveness. The better your tags, the sharper the coaching your agent can deliver.

4. Designing your coaching prompt template

Now it's time to add structured prompts for coaching-style feedback. Design a query template that tells our AI exactly what to evaluate and how to respond. Start by defining the goal: turn this transcript into actionable coaching. Then specify the categories you want analyzed, like tone, objection handling, discovery quality, and closing effectiveness. Set a clear output format, such as two strengths, two improvements, and one key takeaway. This two-two-one structure keeps feedback balanced and digestible. Finally, add tone instructions so your AI matches your coaching style, whether that's encouraging, direct, or analytical. The clearer your prompt template, the more consistent and useful your coaching feedback will be. Remember, we're creating a repeatable coaching framework that can be applied to every call.

5. Interpreting coaching notes with transcript evidence

Here's what our coaching output should look like: specific, evidence-based, and actionable. Each strength and improvement should reference something that actually happened in the call, not generic advice. Feedback should feel like it came from someone who actually listened to the call, because your AI did analyze it, turn by turn. The takeaway ties it all together with one clear action the rep can apply in their very next conversation.

6. Let's practice!

Now it's your turn! Open your Copilot agent and build your coaching prompt. Define your categories, set your format, and test it on a real transcript. Does the feedback feel helpful, human, and ready to use? Refine until it does, and you'll have a coaching assistant that scales your best insights across every call.

Create Your Free Account

By continuing, you accept our Terms of Use, our Privacy Policy and that your data is stored in the USA.