Como Criar Sistema de IA que Analisa Chamadas de Vendas (SEM Código!)

Introduction to Call Monitoring Automation

Overview of the Speaker's Experience

• The speaker shares their background as a sales manager for nearly five years, highlighting their dislike for call monitoring despite its importance in their role.

• They mention that advancements in AI now allow for quicker and more efficient call monitoring processes compared to traditional methods.

Purpose of the Video

• The video aims to demonstrate an automation process using N8N and a dashboard created with Lovable, designed to transcribe, analyze, score, and suggest improvements for sales calls.

• The speaker emphasizes the significant time savings achieved through this automation, which they have previously sold to teams of up to 15 salespeople.

Understanding Call Monitoring

Definition and Importance

• Call monitoring is defined as the evaluation and feedback process for sales calls, crucial for larger sales teams with structured management.

• Managers often spend extensive hours reviewing recorded calls to provide structured feedback or training, making it a vital yet time-consuming task.

Challenges Faced by Sales Managers

• The speaker recounts personal experiences of spending excessive time on call monitoring tasks, sometimes neglecting them due to their tedious nature.

• Despite its challenges, effective call monitoring is essential for team development and continuous improvement in sales performance.

Demonstration of Automation Process

Uploading Calls and Processing Data

• The speaker begins demonstrating the upload process of a new call file while selecting the team member being evaluated. This initiates data processing within the system.

• They explain how N8N automates data handling by receiving information from Lovable and utilizing OpenAI for transcription into text format.

Data Organization and Analysis

• A trained agent evaluates the transcribed calls based on specific prompts designed for effective monitoring; results are stored in Supabase database used by Lovable’s frontend interface.

• After processing is complete, users can view scores along with detailed insights about each call including duration and suggestions for improvement provided by AI analysis.

Dashboard Features

Filtering Capabilities

• The dashboard allows filtering by team members, processing status of calls, and date comparisons; changes reflect immediately across all displayed data points.

• Key metrics such as total calls during a period, average scores received per call, current month statistics, and average duration are prominently displayed on the dashboard interface.

Building the Automation in N8N

Initial Setup Requirements

• The first step involves creating a webhook connected to Lovable that continuously monitors incoming information related to new audio uploads from users’ actions within the system.

Data Handling Within N8N

• Upon receiving new audio files via webhook triggers in N8N, additional nodes are utilized to organize incoming data effectively before further processing occurs within the automation workflow.( t = 317 s)

How to Use OpenAI for Audio Transcription and Data Organization

Utilizing GPT for Code Generation

• The speaker discusses using a technique called "ask GPT" to generate code that effectively organizes data, transforming it into a format suitable for sending to an OpenAI node capable of audio transcription.

Accessing OpenAI Models

• The speaker highlights the availability of various models on OpenAI, specifically mentioning an audio action model designed for transcribing recordings.

Receiving Transcription Results

• Upon sending audio data to the transcription model, users receive a text output along with the duration of the audio in seconds, which is useful for frontend applications.

Crafting Effective Prompts

• A basic prompt example is provided: it instructs the AI to analyze sales call transcriptions and return results in a specific JSON format. This structure includes evaluation criteria such as rapport building and objection handling.

Structuring Data Outputs

• Emphasis is placed on ensuring outputs are formatted correctly (in JSON), facilitating further processing in subsequent nodes like database uploads. The agent organizes this data efficiently.

Integrating with Supabase

• The speaker explains how Supabase can be used as a database integrated with Lovable, allowing automatic table creation and easy access through N8N workflows.

Updating Database Records

• Instructions are given on updating rows in Supabase by mapping conversation IDs and passing relevant scores, transcriptions, feedback, duration, and improvement suggestions from AI agents.

Populating Database Tables

• As calls are processed, tables populate with essential information including call IDs generated by Lovable, timestamps, caller details, full transcriptions, scores, feedback suggestions, statuses, and durations.

Enhancing Frontend Development

• The integration between Lovable and backend systems requires iterative communication to refine prompts for optimal frontend results.

Resources for Prompt Creation

• The speaker mentions providing resources in the video description that include initial prompts developed through extensive interaction with Cloud AI tools.

Prompt for Lovab Integration

Initial Setup and Prompt Usage

• The video introduces a definitive prompt to be used in Lovab, emphasizing that users should copy the provided prompt from the material description for effective results.

• Viewers are advised to replace the webhook in the prompt with their N8N automation webhook to ensure proper functionality within Lovab.

Connecting to Supabase

• After pasting the prompt into Lovab, users are instructed to connect their Supabase account by clicking a designated button, which is crucial for database integration.

• The initial output from Lovab includes instructions on connecting to a database, highlighting its user-friendly guidance throughout the setup process.

Frontend Development Insights

• The first version of the frontend is presented as visually appealing and functional, featuring an upload button and filter modals similar to previous designs but with enhancements.

• Users can begin testing functionalities immediately after connecting with N8N using the same webhook from earlier prompts.

Database Table Creation

• With established connections, users are prompted to request table creation in Supabase without needing specific details; Lovab will handle this automatically through native integration.

• Once tables are created in Supabase, users can verify them via SQL or table editors within their project dashboard.

Managing Frontend and Backend Integration

• A distinction is made between frontend (Lovab interface) and backend (N8N control), stressing that both must be managed concurrently for successful system operation.

Advanced System Features

• The speaker discusses potential enhancements like integrating call recording systems directly into workflows, allowing for automated data collection without manual uploads.

Call to Action and Resources

• Viewers interested in implementing similar systems are encouraged to reach out via contact information provided in the description for personalized assistance.

• Additional resources include downloadable automation code for N8N available in the material description, streamlining setup processes significantly.

Engagement Encouragement

• The video concludes with a reminder for viewers to like and subscribe if they found value in the content shared.