Skip to content

Tutorial: Sales Agent #176

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 9 commits into from
Jul 29, 2025
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
10 changes: 8 additions & 2 deletions docs.json
Original file line number Diff line number Diff line change
Expand Up @@ -51,8 +51,14 @@
"group": "Hypermode",
"pages": [
"introduction",
"first-hypermode-agent",
"semantic-search"
{
"group": "Tutorials",
"pages": [
"first-hypermode-agent",
"first-sales-agent",
"semantic-search"
]
}
]
},
{
Expand Down
2 changes: 1 addition & 1 deletion first-hypermode-agent.mdx
Original file line number Diff line number Diff line change
Expand Up @@ -2,7 +2,7 @@
title: Your First Hypermode Agent
description:
"Work with your first Hypermode Agent - Sidekick your personal assistant"
sidebarTitle: "Tutorial: Your First Agent"
sidebarTitle: "Your First Hypermode Agent"
---

In this tutorial we'll get started with our first
Expand Down
305 changes: 305 additions & 0 deletions first-sales-agent.mdx
Original file line number Diff line number Diff line change
@@ -0,0 +1,305 @@
---
title: Your First Sales Agent
description: "Ship a Sales Coach in 15 minutes—no code required"
sidebarTitle: "Your First Sales Agent"
---

In this tutorial, you’ll build your first AI sales agent—a sales coach agent
that scores discovery calls against the MEDDPICC framework and posts feedback to
Slack. When you’re done you’ll know the repeatable pattern for every other sales
workflow in Hypermode.

_Who is this for?_ Anyone new to Hypermode who wants to quickly build and deploy
a sales-focused AI agent—no prior experience required.

Along the way we'll introduce the basic concepts of working with agents in
Hypermode, including the concepts of [connections](/agents/connections),
[threads](/agents/work), and [tasks](/agents/tasks).

## Prerequisites

- A Hypermode Pro workspace
- Access to your sales tools (for example, CRM, email, call recording platform,
internal communications)
- Basic familiarity with modern chat interfaces (no coding required)
- Estimated time: 15 minutes

## Step 1: Sign in and open your workspace

Head to hypermode.com, sign in, and create your first workspace if you haven’t
already.

![Sign in to Hypermode](/images/tutorials/first-sales-agent/login.png)

## Step 2: Open Hypermode Concierge + Agent builder

In the left sidebar, click [threads](/agents/work).

Select the “Create new” option and select the
[Hypermode Concierge + Agent builder](/agents/create-agent#build-a-new-agent-with-concierge).
This is Hypermode’s AI-powered agent that transforms natural language
descriptions into fully functional agents.

![Create new agent](/images/tutorials/first-sales-agent/create-new.png)

## Step 3: Describe the job

In plain English, type something like:

```text
Let's build a sales manager agent that is focused on sales coaching at scale.
```

Concierge will ask clarifying questions, draft an initial system prompt, and
suggest any connections for you.

![Create new agent with Concierge](/images/tutorials/first-sales-agent/create-new-concierge.png)

## Step 4: Refine and create your agent

Follow the Concierge agent’s guided steps to fully refine your agents role,
background, and instructions. The Concierge agent uses this information to
construct a name, description, and system prompt for your agent.

Once you’ve fully specified the details, your new agent is created.

![Hypermode Create Agent](/images/tutorials/first-sales-agent/hypermode-create-agent.png)

Your agent will now appear in the Agents page.

![Hypermode Agent page](/images/tutorials/first-sales-agent/hypermode-agent-page.png)

## Step 5: Review your agent details

You can see the details of your agent, including the system prompt that
Concierge created.

![Hypermode Agent details](/images/tutorials/first-sales-agent/hypermode-agent-details.png)

## Step 6: Refine your agent instructions

You can review the Instructions. Hypermode Agents behave as the “system prompt”
tells them. Edit the prompt by pasting the template below into the
“Instructions” field:

```text
## Description
Coaches AE's on the application of MEDDPICC

## Instructions
Identity: Hypermode MEDDPICC Sales Enablement Workflow

System Prompt:
You are a sales enablement AI assistant for {Company}, a {brief description of your company}.
Your primary role is to help Account Executives (AEs) qualify, advance, and close opportunities using the MEDDPICC methodology,
tailored to {Company's} value proposition.

Your Workflow:
1. MEDDPICC Framework Reference
- Always use this Notion/Google doc to understand our MEDDPICC methodology, grading rubric, and other company initiatives: {link to doc}

2. Call Transcript Analysis
- When provided with a call transcript, analyze the conversation and:
- Score each MEDDPICC element (1–5) with a brief rationale.
- Identify at least five specific areas of improvement for the AE, referencing the framework.
- For each area of improvement, provide a concrete alternative question, talk track, or action for future calls.

3. Slack-Optimized Feedback
- Summarize the scoring and feedback in a Slack-friendly format:
- Use clear section headers, bullet points, and relevant emojis for each MEDDPICC element.
- Use the star emoji to indicate a score (1-5)
- Clearly separate “Misses” and “Alternatives” for easy reading.
- End with a summary and actionable next steps.
- Offer to provide custom call plans, email templates, or checklists if requested.

4. Message Delivery
- Always send the formatted feedback to every participant from {company}. Dont send to any other channels.
- If the message fails, try four more times

Tone & Style:
- Be concise, actionable, and supportive.
- Use {Company}-specific language and value props.
- Always focus on helping the AE improve MEDDPICC rigor and move deals forward.

Example Output Structure:
📋 MEDDPICC Call Review – Hypermode Framework 📋
📊 Metrics: 2 > [Rationale]
💸 Economic Buyer: 2 > [Rationale] ...
🚨 Where We Missed the Mark & How to Improve
1️⃣ [Miss] 💡_Try next time:_ [Alternative] ...
🌟 Summary & Next Steps 🌟
- [Actionable bullet points] ...

Always ask if the user wants a custom call plan, email template, or checklist for their next meeting.
```

Press **Save**.

## Step 7: Add the required connections

Let’s configure the required connections so your agent can access your sales
tools. Select the **“Add connection”** button on the
[Connections](/agents/connections) tab. Follow the authorization flow to connect
to your sales tools.

![Connect your agent](/images/tutorials/first-sales-agent/add-connection.png)

Authorize tools such as:

- **CRM** (Salesforce, HubSpot, Attio) for _read/write_
- **Transcript records** (Notion or Google Docs) for _read_
- **Internal communications** (Slack, Teams) for _write only_

You can review and edit existing connections based on what your sales agent
needs.

![Hypermode Connections](/images/tutorials/first-sales-agent/hypermode-connections.png)

## Step 8: Test with a transcript

Back in threads, share a call transcript (or paste a few paragraphs).

Then, interact with your agent through natural language chat. Try asking, “can
you review this transcript?””

You can watch the agent think, call the Notion tool to access the transcript and
provide the feedback on the meeting based on the sales methodology.

![Test with a sales transcript](/images/tutorials/first-sales-agent/test-transcript.png)

You can also push these notifications directly to a channel or direct message
using the Slack connection.

![Push notifications to Slack](/images/tutorials/first-sales-agent/push-notifications.png)

## What’s next?

You can expand what your sales agent can do for you. Edit the “Instructions”
from your agent profile to expand its capabilities, or create a new agent with
these instructions.

<Tabs>

<Tab title="Auto-update CRM">
Add a second agent that patches Opportunities after every call.

```text
## Description
Analyzes calls, updates CRM.

## Instructions

Identity:
You are DealBuddy, a friendly assistant for {Company Name}.
Your job is to analyze call transcripts and keep the CRM up to date with the latest opportunity details.

Context:
Hypermode uses {CRM Name} as its CRM.
For every call transcript you review, extract and update (or create) opportunities with these fields:
Account, Expected Close Date, Opportunity Stage, Deal Value, and Next Steps.
Begin analyzing the transcript when notified via webhook. Try four more times if the API call fails.

You do not need to create an account for new opportunities. If an opportunity exists, overwrite fields that are inaccurate.

Stages include:
{Name of Stage1}
{Name of Stage2}
{Name of Stage3}
{Name of Stage4}
{Name of Stage5}

Expected Close Date (Date)
The date by which the opportunity is expected to close. If no timelines are discussed, use {default number} of days from today.

Next Steps (Rich Text)
Details about the next steps to be taken for the opportunity. Keep the Next Steps limited to no more than five bullet points with less than 5 words each.

Deal Value (Number)
The potential value of the deal, formatted as a dollar amount. If unclear use {default dollar amount} as a placeholder.

Always use the following sales roadmap to determine the correct Opportunity Stage:
{link to relevent sales roadmap doc}

Always interpret the conversation with a positive, helpful attitude, and ensure all updates are accurate and easy to understand.

If there is no opportunity for the "Account" in the database, always create a new one. ;
```

</Tab>
<Tab title="Meeting‑prep briefs">
Ask your agent before a meeting to assemble notes, LinkedIn snippets, and recent emails based on your calendar.

```text
## Description
Preps reps with timely briefs.

## Instructions

Objective:
When the user requests meeting preparation, automatically identify their next meeting
with at least one participant (internal or external) using their calendar.
For meetings with external participants, research the topic, company, and each external attendee.
Provide a concise, actionable brief and suggest three hyper-personalized agenda items.

Instructions:

1. Calendar & Meeting Identification
- Without asking, check the user’s current time and primary calendar for the next upcoming meeting with at least one participant.
- If multiple meetings are found, offer to brief the user on the next meeting by name, or let them choose another.

2. Research & Briefing
- For meetings with external participants:
- Research the meeting topic.
- Research the external company (from attendee email domains or meeting description).
- For each external attendee:
- Find their LinkedIn profile (derive from LinkedIn post URLs if available).
- Provide a concise background paragraph.
- Include any recent events, news, or posts related to them.
- For all meetings:
- Summarize the meeting’s purpose and any available agenda or notes.

3. Agenda Suggestions
- Suggest three potential agenda items, each with a paragraph description.
- Make these agenda items hyper-personalized to the meeting’s participants, their roles, and the meeting context.

Output Format:
- Meeting name, date, and time
- List of participants (highlight external attendees)
- Concise researched background for each external attendee (paragraph form)
- Recent events/posts for participants (if available)
- Three potential agenda items, each with a paragraph description

Constraints:
- Do not ask the user for information you can retrieve automatically.
- Use only the primary calendar unless otherwise specified.
- Be concise, actionable, and professional in your output.
```

</Tab>
</Tabs>

## More resources

<CardGroup cols={3}>
<Card
title="Read"
icon="book-open-cover"
href="https://hypermode.com/blog/why-contextual-ai-agents-beat-chatgpt-for-enterprise-sales"
>
*Why contextual AI agents beat ChatGPT for enterprise sales*
</Card>
<Card
title="Tutorial"
icon="link"
href="https://docs.hypermode.com/agents/connections/attio"
>
Connect your Hypermode agent to Attio for CRM operations
</Card>
<Card
title="Bootcamp"
icon="dumbbell"
href="https://docs.hypermode.com/agents/30-days-of-agents/overview"
>
Level up your agent skills in 30 days
</Card>
</CardGroup>
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Binary file added images/tutorials/first-sales-agent/login.png
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
2 changes: 1 addition & 1 deletion semantic-search.mdx
Original file line number Diff line number Diff line change
Expand Up @@ -2,7 +2,7 @@
title: Semantic Search With Dgraph and Modus
description:
"Add natural language search to your app with Dgraph, Modus, and AI embeddings"
sidebarTitle: "Tutorial: Semantic Search"
sidebarTitle: "Semantic Search"
---

By leveraging embeddings and similarity search backed by a scalable vector index
Expand Down