Build a Multi-Channel Sales Agent (No-Code)

How to Build a No-Code Sales Agent

Introduction to the No-Code Sales Agent

• The video introduces a no-code sales agent capable of researching leads, making personalized calls, qualifying leads over the phone, updating CRM systems, and sending follow-ups via WhatsApp, email, and LinkedIn—all on autopilot.

• The presenter, Ben, shares that he created this system for his inbound sales funnel but emphasizes its applicability for outbound sales as well. A demo and step-by-step breakdown will be provided.

Overview of the System Setup

• Ben explains that the AI agent is triggered by a lead form on his website where potential clients provide information about their needs.

• He demonstrates filling out an example lead form with vague answers regarding sales automation projects to illustrate how the system works.

Lead Research and Qualification Process

• Upon submission of the lead form, the AI agent begins researching both the lead and their company to create a personalized call script.

• The AI then makes a voice call using this script to gather more information about the project and qualify the lead based on budget discussions.

Example Call Demonstration

• During a simulated call with "Brian," Ben's AI assistant asks questions about specific challenges in implementing automation services.

• Brian mentions interest in an outbound email system for personalized outreach and marketing automation for content creation on LinkedIn.

Follow-Up Actions Post Call

• After qualifying Brian as a suitable lead based on budget alignment, Ben’s AI informs him that further communication will occur shortly.

• Following up through email and WhatsApp occurs automatically after qualification; messages include details discussed during the call along with scheduling links for future meetings.

Conclusion of System Functionality

• The video concludes with an overview of how follow-up messages are sent via different platforms while noting limitations when contacting high-profile accounts like Brian Chesky from Airbnb.

How to Set Up an AI Agent System

Introduction to Relevance AI

• The speaker introduces the concept of setting up a system using Relevance AI, a no-code AI agent builder that allows users to create powerful agent solutions easily.

• A link will be provided in the description for additional tutorials on building AI agents and access to a free template for this specific agent solution.

Integration with Make.com

• The speaker explains the use of make.com to trigger the agent from Typeform due to its native integration, simplifying the process compared to using custom webhooks in Relevance AI.

Workflow Overview

• The workflow begins when a Typeform is filled out, triggering the sales agent who receives lead information and starts processing it.

• The sales agent uses various tools: first, it finds LinkedIn profiles of leads and companies; then it scrapes relevant data from these profiles for summaries.

Sales Agent Functionality

• After gathering information, the sales agent can initiate voice calls with leads using tool three, generating personalized call scripts based on collected data.

• Depending on whether leads are qualified or not, different follow-up actions are taken—sending WhatsApp messages or emails with booking links if qualified.

Handling Unqualified Leads

• If a lead is deemed unqualified or has budget constraints, the sales agent can refer them to partner agencies that may offer more suitable services.

CRM Updates and Conclusion

• The final step involves updating the CRM with all gathered information from both tools and call results. This setup is described as straightforward yet effective.

Step-by-Step Setup in Relevance AI

Demonstration of Process

• The speaker demonstrates how the system works within their Relevance AI account by showing completed processes triggered by lead forms.

Agent Prompt Configuration

• An overview of how triggers work in Relevance AI is provided; since there’s no direct Typeform trigger available, make.com is utilized instead.

Core Instructions for Agents

• Details about configuring core instructions for agents are shared. These include defining roles such as "world-class inbound lead agent" focused on researching leads and engaging through multiple channels.

Goals and Objectives

Understanding the Role of AI Agents in Business Automation

The Importance of Context in AI Automation

• Providing context is crucial for language models and agents to understand their tasks within a larger framework, enhancing their effectiveness in business automation.

• Clear Standard Operating Procedures (SOPs) are essential for guiding agents through step-by-step actions, especially when multiple tools are involved.

Tool Utilization and Instructions

• Specific instructions must be given regarding tool usage; for instance, only using the "send LinkedIn invite" tool if a relevant profile is found.

• Detailed descriptions of each tool's purpose and application help agents understand how to utilize them effectively.

Examples and Scenarios

• Providing examples helps agents navigate different scenarios by illustrating potential actions based on varying inputs.

• An AI agent prompting tool can assist in generating these examples quickly, improving the reliability of agent responses.

Enhancing Agent Reliability

• Reinforcing important rules through notes ensures that agents adhere to critical guidelines during operations.

• Flow Builders can simplify complex processes by visually mapping out conditions and actions, thereby increasing agent reliability across various scenarios.

Managing Lead Qualification

• Labeling leads as qualified or unqualified allows for better organization and tracking within the system.

• Tools can be set to auto-run or require human approval depending on the use case, allowing flexibility in managing lead interactions.

Understanding Tool Functionality

• Tools operate based on logic sequences that dictate task performance; understanding this structure is vital for effective implementation.

• Inputs such as names and company URLs are used to generate search queries that facilitate finding leads' LinkedIn profiles.

LinkedIn Scraping and Automation Process

Overview of LinkedIn Data Retrieval

• The process begins with a prompt to find the LinkedIn profile of Brian Chesky, co-founder of Airbnb. An example is provided in a free template for clarity.

• The output, which includes the LinkedIn URL, is stored in a variable that is then passed into an integrated LinkedIn scraper within the AI system.

• A condition checks if the LinkedIn profile was found; if not, it outputs "LinkedIn not available," preventing further steps from executing.

Data Extraction and Summary Creation

• Upon successful scraping, extensive data from the user's LinkedIn profile is retrieved. This data undergoes an AI-driven summarization to extract key information.

• Outputs include lead summaries and both company and lead LinkedIn URLs. Users can customize outputs by setting parameters manually to receive multiple results.

Company Research Tool Implementation

• The next step involves using a research tool where the previously found company LinkedIn URL and website URL are inputted for further analysis.

• A contingency plan is established for scenarios where no company LinkedIn URL is found; it outputs "company LinkedIn URL not available" to avoid system errors.

Web Scraping Techniques

• If the company’s LinkedIn URL isn't available, recursive web scraping techniques are employed to attempt multiple data retrieval attempts with delays between each try.

• This method ensures robust data collection from the company's website while generating a comprehensive summary based on scraped information.

Voice Call Preparation

• After gathering all necessary information about leads and companies, this data feeds into a voice call tool designed for outreach purposes.

• Specific instructions are given to ensure accurate filling out of fields by agents, emphasizing clarity in descriptions for effective automation.

Personalization of Outreach Scripts

• Key variables such as first name, lead summary, company summary, and phone number are compiled. Formatting requirements (e.g., including "+" before phone numbers) are clearly communicated to agents.

• A personalized line script generator creates engaging introductions based on context provided by previous inputs. Examples guide agents in crafting tailored messages.

Execution of Calls

• The personalized line generated serves as part of the call script used during outreach efforts. This approach aims at enhancing engagement through customized communication strategies.

AI-Driven Voice Assistant Development

Balancing Personalization and Guidelines

• The integration of AI in voice assistants aims to combine personalization with established guidelines, enhancing reliability. A structured template is essential for maintaining quality.

Built-in Relevance AI Features

• The system utilizes built-in relevance AI, eliminating the need for third-party software like VAP or Sylow, streamlining the process of creating voice agents.

Customizing Prompts for Leads

• Personalization of prompts based on specific leads is crucial. This setup simplifies the integration process compared to building from scratch in other platforms.

Importance of Scripted Objectives

• Each call should have clear objectives: gather information about desired automations, communicate pricing, and inform leads that Ben will follow up shortly.

Mandatory Script Elements

• Certain lines in the script are mandatory and must be recited verbatim to ensure consistency and adherence to guidelines. Flexibility can be allowed but within defined limits.

Tone of Voice Considerations

• Establishing a friendly yet professional tone is vital for making conversations feel natural while ensuring respectfulness and informativeness.

Handling User Inquiries Effectively

• Instructions on how to respond to common user questions (e.g., inquiries about being an AI or talking to a human) should be included in the prompt for better engagement.

Limitations of Current Setup

• The current setup lacks a knowledge base feature that allows real-time question answering during calls, which could enhance user interaction significantly.

Recommendations for Enhanced Capabilities

• For more advanced functionalities such as tool calling during conversations (e.g., booking meetings), it’s advisable to consider alternative setups like VAPY or Synf Flow.

Call Details Management

AI-Driven Call Analysis and Follow-Up Automation

Overview of Call Processing

• The tool processes call details, extracting specific information from transcripts to determine the call status (successful or unsuccessful).

• Success is defined not just by whether the call was answered but also if a meaningful conversation occurred.

• Key outputs include call status, automation needs discussed during the call, and qualification status based on budget questions.

Data Extraction and Agent Notification

• Extracted data includes a summary of the call, recording, transcript, and qualification status which are sent back to the agent for further action.

Email Automation Based on Call Outcomes

• An email is generated based on qualification status (qualified/unqualified), call success, and other relevant details.

• Personalized messages are crafted depending on whether the call was successful or not.

WhatsApp Integration for Communication

• The integration allows sending personalized messages via WhatsApp without needing the official business API setup.

• A simple connection process involves scanning a QR code to link personal WhatsApp accounts with the system.

LinkedIn Invitation Tool Functionality

• The tool enables sending LinkedIn invites using usernames rather than full URLs; personalization is key in these messages.

How to Effectively Use LinkedIn and CRM Tools

Connecting on LinkedIn

• The speaker demonstrates sending a connection request on LinkedIn, confirming that the action was successful.

• A personal message was also sent along with the connection request, indicating an effort to engage meaningfully.

• The process is described as straightforward, emphasizing the effectiveness of using LinkedIn for networking.

Updating CRM with HubSpot

• The speaker outlines how to fill out various fields in HubSpot's CRM tool, including company summary and lead details.

• Key information includes job title, LinkedIn URL, company website URL, industry specifics, and lead score.

• Emphasis is placed on utilizing the HubSpot API call module for creating new contacts within the system.

Integration Options

• For those intimidated by direct API integration, an alternative method through make.com is suggested for easier connectivity with HubSpot.

• The speaker reassures viewers that while it may seem complex at first glance, using these tools is not overly difficult once familiarized.

Conclusion and Engagement