Neo AI-Powered WhatsApp Running Coach Development by Kotiku TaiyeNeo AI-Powered WhatsApp Running Coach Development by Kotiku Taiye

Neo AI-Powered WhatsApp Running Coach Development

Kotiku Taiye

Completed work

AI Agent Designer

AI Automation

Fullstack Engineer

N8N

Stripe

Supabase

Electronics

🏃‍♂️ Neo - AI-Powered WhatsApp Running Coach

Full-stack subscription-based training platform delivering personalized coaching through WhatsApp using n8n, OpenAI GPT-4o, Stripe, and Supabase.

📋 Table of Contents

🎯 Overview

Neo transforms professional running coaching into an accessible WhatsApp experience. Traditional coaches cost €100-300/month and require constant availability. Neo provides 24/7 AI-powered coaching at €9.99/month through automated subscription management and personalized training plans.

Problem: Running coaches are expensive and inaccessible. Solution: AI-powered WhatsApp bot with automated plan generation, progress tracking, and subscription billing. Result: Scalable coaching platform handling unlimited users with zero manual intervention.

✨ Features

🤖 Conversational Onboarding

Natural language data collection via WhatsApp

User profiling (age, weight, running level, goals, availability)

Instant personalized plan generation

💳 Subscription Management

Three pricing tiers (€9.99/month, €55/6-months, €100/year)

Stripe Checkout integration

Discount code system

Automated subscription lifecycle via webhooks

📅 Automated Weekly Plans

Every Monday at 8 AM generation

Progress-based adaptation (completion rate + difficulty analysis)

Personalized sessions with distance, pace, and nutrition tips

🏋️ Workout Tracking

Session logging (distance, time, pace, difficulty)

AI-powered tool for data extraction

Performance history and analytics

⏰ Daily Reminders

Smart scheduling based on weekly plan

Completion verification

Motivational messaging

💾 Persistent Memory

100-message chat history

Weekly plan storage

Workout history tracking

Context-aware responses across sessions

🏗️ Architecture

Neo consists of 6 interconnected n8n workflows:

1. Master Router

Intelligent message routing with 3 paths:

Onboarding Route: New user data collection

Payment Route: Stripe checkout flow

Premium Route: Active subscriber interactions

2. Weekly Plan Generator

Automated Monday morning execution:

Fetches active users

Analyzes previous week performance

Generates personalized plans via GPT-4o

Saves to database and sends via WhatsApp

3. Workout Reminders

Daily 9 PM check-in:

Identifies users with scheduled workouts

Checks completion status

Sends motivational reminders

4. Stripe Webhook Handler

Subscription lifecycle automation:

Subscription created/updated/canceled events

Database status synchronization

Welcome/cancellation messaging

5. Premium Agent

AI-powered coaching with custom tools:

GPT-4o integration

Workout completion tool

Chat history persistence

Context-aware responses

6. Database Operations

Persistent storage layer:

User profiles

Chat history (100 messages)

Weekly plans

Workout tracking

🛠️ Tech Stack

Category Technology Workflow Automation n8n Messaging WhatsApp Business API AI/ML OpenAI Database Supabase (PostgreSQL) Payments Stripe Architecture Webhook-based, Event-driven

📸 Screenshots

User Experience

Conversational Onboarding

Personalized Weekly Plan

Stripe Subscription Checkout

Backend Architecture

Master Router Workflow

🔧 Key Challenges & Solutions

Challenge 1: Webhook Routing Conflicts

Problem: Multiple WhatsApp triggers caused duplicate message handling.

Solution: Implemented master router pattern with intelligent 3-route architecture, consolidating all WhatsApp webhooks into a single entry point.

Impact: Eliminated duplicate executions, improved reliability.

Challenge 2: Data Persistence

Problem: WhatsApp conversation memory expired after 24 hours, losing user context.

Solution: Built database-backed persistence layer storing chat history, user profiles, and workout data.

Impact: Seamless conversations with full context across sessions.

Challenge 3: Stripe Integration

Problem: Checkout sessions failing due to Product ID vs Price ID confusion.

Solution: Corrected implementation to use Stripe Price IDs with proper metadata tracking.

Impact: 100% successful checkout conversion.

Challenge 4: AI Plan Quality

Problem: Generic plans without progression logic.

Solution: Created progression engine analyzing completion rate (0-100%), difficulty ratings (1-10), and workout history to adapt intensity.

Logic:

High difficulty (8+) → Reduce intensity 10-15%

Low difficulty (≤3) + 80% completion → Increase load 5-10%

Low completion (<50%) → Maintain intensity, increase motivation

Impact: Truly personalized, adaptive training plans.

Challenge 5: Double Execution

Problem: Multiple GET nodes connected to single analysis node caused duplicate executions.

Solution: Added Merge nodes with "Merge By Position" mode.

Impact: Prevented duplicate plan generation and database writes.

💾 Database Schema

-- Users Table
CREATE TABLE users (
phone TEXT PRIMARY KEY,
nom TEXT,
edat INTEGER,
pes DECIMAL,
sexe TEXT,
terreny TEXT,
dies_disponibles INTEGER,
disponibilitat_detallada TEXT,
nivell TEXT,
objectius TEXT,
subscripcio TEXT CHECK (subscripcio IN ('active', 'trial', 'canceled')),
stripe_customer_id TEXT,
onboarding_completed BOOLEAN DEFAULT false,
created_at TIMESTAMPTZ DEFAULT NOW()
);-- Chat History Table
CREATE TABLE chat_history (
id UUID PRIMARY KEY DEFAULT gen_random_uuid(),
user_phone TEXT REFERENCES users(phone),
role TEXT CHECK (role IN ('user', 'assistant')),
message TEXT,
created_at TIMESTAMPTZ DEFAULT NOW()
);-- Weekly Plans Table
CREATE TABLE weekly_plans (
id UUID PRIMARY KEY DEFAULT gen_random_uuid(),
user_phone TEXT REFERENCES users(phone),
week_number INTEGER,
plan_text TEXT,
status TEXT CHECK (status IN ('active', 'completed')),
created_at TIMESTAMPTZ DEFAULT NOW(),
UNIQUE(user_phone, week_number)
);-- Workout History Table
CREATE TABLE workout_history (
id UUID PRIMARY KEY DEFAULT gen_random_uuid(),
user_phone TEXT REFERENCES users(phone),
week_number INTEGER,
session_number INTEGER,
distance DECIMAL(5,2),
time_minutes DECIMAL(6,2),
pace_min_per_km DECIMAL(5,2),
difficulty_rating INTEGER CHECK (difficulty_rating BETWEEN 1 AND 10),
feedback TEXT,
completed BOOLEAN DEFAULT true,
completion_date TIMESTAMPTZ,
created_at TIMESTAMPTZ DEFAULT NOW(),
UNIQUE(user_phone, week_number, session_number)
);

💰 Business Model

Tier Price Discount Features Monthly €9.99/mo - Pay as you go 6-Month €55 8% Best value for commitment Annual €100 16% Maximum savings

Free Trial: 7-day trial period on all plans

Discount Codes: Support for promotional codes

📦 Installation

Note: This is a portfolio project. The workflows are provided for educational purposes. Running this system requires:

n8n instance

WhatsApp Business API account

OpenAI API key

Stripe account

Supabase database

Prerequisites

n8n (self-hosted or cloud) WhatsApp Business API credentials OpenAI API key Stripe API keys (test + live) Supabase account

Quick Start

Clone Repository

git clone https://github.com/taiye-kotiku/neo-ai-running-coach.git
cd neo-ai-running-coach

Import Workflows

Import JSON files from /workflows directory into n8n

Update credentials for all integrations

Setup Database Run schema.sql in Supabase SQL Editor

Configure Webhooks

Set WhatsApp webhook to Master Router URL

Set Stripe webhook to Webhook Handler URL

Update Environment Variables

Add API keys to n8n credentials

Update phone numbers and Stripe Price IDs

See SETUP.md for detailed instructions.

📊 Metrics📈 System Scale:

6 production workflows 4 API integrations 4 database tables 100-message chat memory Unlimited concurrent users Sub-second response times

🚀 Future Enhancements

Mobile app integration

Heart rate zone monitoring

Race preparation programs

Nutrition tracking

Community features

Multi-language support

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

📧 Contact

Built by: Taiye Promise Kotiku

Portfolio: [https://taiye.framer.website] LinkedIn: [www.linkedin.com/in/taiye-kotiku] Email: [promisetaiye16@gmail.com | kotikutaiye@gmail.com] Twitter: [@taiyekotiku]

Interested in building something similar? I'm available for freelance AI automation projects! Let's talk →

⭐ If you found this project interesting, please star it!

Like this project

Completed work

Posted Jan 1, 2026

Developed an AI-based running coach platform using n8n, OpenAI, and other tech tools.

Likes

Views

Neo AI-Powered WhatsApp Running Coach Development

Kotiku Taiye

Completed work

AI Agent Designer

AI Automation

Fullstack Engineer

N8N

Stripe

Supabase

Electronics

🏃‍♂️ Neo - AI-Powered WhatsApp Running Coach

Full-stack subscription-based training platform delivering personalized coaching through WhatsApp using n8n, OpenAI GPT-4o, Stripe, and Supabase.

📋 Table of Contents

🎯 Overview

✨ Features

🤖 Conversational Onboarding

Natural language data collection via WhatsApp

User profiling (age, weight, running level, goals, availability)

Instant personalized plan generation

💳 Subscription Management

Three pricing tiers (€9.99/month, €55/6-months, €100/year)

Stripe Checkout integration

Discount code system

Automated subscription lifecycle via webhooks

📅 Automated Weekly Plans

Every Monday at 8 AM generation

Progress-based adaptation (completion rate + difficulty analysis)

Personalized sessions with distance, pace, and nutrition tips

🏋️ Workout Tracking

Session logging (distance, time, pace, difficulty)

AI-powered tool for data extraction

Performance history and analytics

⏰ Daily Reminders

Smart scheduling based on weekly plan

Completion verification

Motivational messaging

💾 Persistent Memory

100-message chat history

Weekly plan storage

Workout history tracking

Context-aware responses across sessions

🏗️ Architecture

Neo consists of 6 interconnected n8n workflows:

1. Master Router

Intelligent message routing with 3 paths:

Onboarding Route: New user data collection

Payment Route: Stripe checkout flow

Premium Route: Active subscriber interactions

2. Weekly Plan Generator

Automated Monday morning execution:

Fetches active users

Analyzes previous week performance

Generates personalized plans via GPT-4o

Saves to database and sends via WhatsApp

3. Workout Reminders

Daily 9 PM check-in:

Identifies users with scheduled workouts

Checks completion status

Sends motivational reminders

4. Stripe Webhook Handler

Subscription lifecycle automation:

Subscription created/updated/canceled events

Database status synchronization

Welcome/cancellation messaging

5. Premium Agent

AI-powered coaching with custom tools:

GPT-4o integration

Workout completion tool

Chat history persistence

Context-aware responses

6. Database Operations

Persistent storage layer:

User profiles

Chat history (100 messages)

Weekly plans

Workout tracking

🛠️ Tech Stack

📸 Screenshots

User Experience

Conversational Onboarding

Personalized Weekly Plan

Stripe Subscription Checkout

Backend Architecture

Master Router Workflow

🔧 Key Challenges & Solutions

Challenge 1: Webhook Routing Conflicts

Problem: Multiple WhatsApp triggers caused duplicate message handling.

Solution: Implemented master router pattern with intelligent 3-route architecture, consolidating all WhatsApp webhooks into a single entry point.

Impact: Eliminated duplicate executions, improved reliability.

Challenge 2: Data Persistence

Problem: WhatsApp conversation memory expired after 24 hours, losing user context.

Solution: Built database-backed persistence layer storing chat history, user profiles, and workout data.

Impact: Seamless conversations with full context across sessions.

Challenge 3: Stripe Integration

Problem: Checkout sessions failing due to Product ID vs Price ID confusion.

Solution: Corrected implementation to use Stripe Price IDs with proper metadata tracking.

Impact: 100% successful checkout conversion.

Challenge 4: AI Plan Quality

Problem: Generic plans without progression logic.

Solution: Created progression engine analyzing completion rate (0-100%), difficulty ratings (1-10), and workout history to adapt intensity.

Logic:

High difficulty (8+) → Reduce intensity 10-15%

Low difficulty (≤3) + 80% completion → Increase load 5-10%

Low completion (<50%) → Maintain intensity, increase motivation

Impact: Truly personalized, adaptive training plans.

Challenge 5: Double Execution

Problem: Multiple GET nodes connected to single analysis node caused duplicate executions.

Solution: Added Merge nodes with "Merge By Position" mode.

Impact: Prevented duplicate plan generation and database writes.

💾 Database Schema

-- Users Table
CREATE TABLE users (
phone TEXT PRIMARY KEY,
nom TEXT,
edat INTEGER,
pes DECIMAL,
sexe TEXT,
terreny TEXT,
dies_disponibles INTEGER,
disponibilitat_detallada TEXT,
nivell TEXT,
objectius TEXT,
subscripcio TEXT CHECK (subscripcio IN ('active', 'trial', 'canceled')),
stripe_customer_id TEXT,
onboarding_completed BOOLEAN DEFAULT false,
created_at TIMESTAMPTZ DEFAULT NOW()
);-- Chat History Table
CREATE TABLE chat_history (
id UUID PRIMARY KEY DEFAULT gen_random_uuid(),
user_phone TEXT REFERENCES users(phone),
role TEXT CHECK (role IN ('user', 'assistant')),
message TEXT,
created_at TIMESTAMPTZ DEFAULT NOW()
);-- Weekly Plans Table
CREATE TABLE weekly_plans (
id UUID PRIMARY KEY DEFAULT gen_random_uuid(),
user_phone TEXT REFERENCES users(phone),
week_number INTEGER,
plan_text TEXT,
status TEXT CHECK (status IN ('active', 'completed')),
created_at TIMESTAMPTZ DEFAULT NOW(),
UNIQUE(user_phone, week_number)
);-- Workout History Table
CREATE TABLE workout_history (
id UUID PRIMARY KEY DEFAULT gen_random_uuid(),
user_phone TEXT REFERENCES users(phone),
week_number INTEGER,
session_number INTEGER,
distance DECIMAL(5,2),
time_minutes DECIMAL(6,2),
pace_min_per_km DECIMAL(5,2),
difficulty_rating INTEGER CHECK (difficulty_rating BETWEEN 1 AND 10),
feedback TEXT,
completed BOOLEAN DEFAULT true,
completion_date TIMESTAMPTZ,
created_at TIMESTAMPTZ DEFAULT NOW(),
UNIQUE(user_phone, week_number, session_number)
);

💰 Business Model

Tier Price Discount Features Monthly €9.99/mo - Pay as you go 6-Month €55 8% Best value for commitment Annual €100 16% Maximum savings

Free Trial: 7-day trial period on all plans

Discount Codes: Support for promotional codes

📦 Installation

Note: This is a portfolio project. The workflows are provided for educational purposes. Running this system requires:

n8n instance

WhatsApp Business API account

OpenAI API key

Stripe account

Supabase database

Prerequisites

n8n (self-hosted or cloud) WhatsApp Business API credentials OpenAI API key Stripe API keys (test + live) Supabase account

Quick Start

Clone Repository

git clone https://github.com/taiye-kotiku/neo-ai-running-coach.git
cd neo-ai-running-coach

Import Workflows

Import JSON files from /workflows directory into n8n

Update credentials for all integrations

Setup Database Run schema.sql in Supabase SQL Editor

Configure Webhooks

Set WhatsApp webhook to Master Router URL

Set Stripe webhook to Webhook Handler URL

Update Environment Variables

Add API keys to n8n credentials

Update phone numbers and Stripe Price IDs

See SETUP.md for detailed instructions.

📊 Metrics📈 System Scale:

6 production workflows 4 API integrations 4 database tables 100-message chat memory Unlimited concurrent users Sub-second response times

🚀 Future Enhancements

Mobile app integration

Heart rate zone monitoring

Race preparation programs

Nutrition tracking

Community features

Multi-language support

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

📧 Contact

Built by: Taiye Promise Kotiku

Portfolio: [https://taiye.framer.website] LinkedIn: [www.linkedin.com/in/taiye-kotiku] Email: [promisetaiye16@gmail.com | kotikutaiye@gmail.com] Twitter: [@taiyekotiku]

Interested in building something similar? I'm available for freelance AI automation projects! Let's talk →

⭐ If you found this project interesting, please star it!

Like this project

Completed work

Posted Jan 1, 2026

Developed an AI-based running coach platform using n8n, OpenAI, and other tech tools.

Likes

Views