The Army of Robots

Building a Modular GPT Direct Messaging System

Case Study

The "Army of Robots" is my custom-built GPT direct messaging system — a fully modular, scalable chat environment featuring custom AI bots, voice integration, image analysis, and dynamic UI. Originally built to support my Choose Your Fighter project, it has now become a mission-critical part of my everyday life. Whether I'm writing songs, cataloging inventory for a retail store, formatting code for my websites, or brainstorming ideas, my Army of Bots powers it all.

Overview

Goals

• Build a modular, easily expandable GPT direct message system.
• Integrate ElevenLabs voice capabilities for natural speech playback.
• Implement GPT-4o Vision analysis to allow bots to "see" uploaded images.
• Develop a clean, stylish dark-mode UI that feels personal and inviting.
• Make new bot creation as fast, streamlined, and scalable as possible.

What I Built

Modular Bot Architecture

Each bot lives in its own PHP + JS page, powered by a shared logic system. Adding a new bot takes just minutes: duplicate the template, update the systemContext and assistantContext, drop in a new avatar, and the bot is live.

AI Voice Playback

Bots are fully voice-enabled using ElevenLabs API. With a single click, any bot can read its replies out loud with near-zero latency, making conversations feel far more dynamic and real.

Dynamic Friends List

The left sidebar dynamically displays all available bots. Clicking a bot instantly switches conversations and loads the new context, all without needing to reload the entire page.

Image Upload and Vision Analysis

Users can upload images directly into the chat. Bots analyze these images using GPT-4o Vision models, enabling them to react to screenshots, diagrams, designs, or any visual content shared mid-conversation.

Custom Styling and UX

The entire system was built from scratch with custom CSS, focusing on a smooth, dark-mode aesthetic with glow effects, animated elements, and mobile responsiveness. Every bot has its own custom avatar and personality styling.

Behind the Scenes: How It Works

• Bot Personalities: Each bot is defined using modular systemContext and assistantContext blocks, making personality swapping or updates frictionless.
• Global JS Core: A single, powerful general.js script handles GPT messaging, markdown rendering, voice playback, image uploading, and lightbox viewing across all bots.
• UI Architecture: Two-panel design — Friends List on the left, Conversation on the right — with full scrollable message history and typing indicators for realism.

Technologies Used

• HTML + PHP: Modular page architecture for each bot.
• JavaScript: Core chat functions, voice integration, dynamic loading, image preview and analysis.
• OpenAI API (GPT-4o Chat + Vision): For conversation and image analysis.
• ElevenLabs API: For real-time text-to-speech voice playback.
• Custom CSS: Full dark-mode support, responsive layout, and animated UI elements.

Why It Matters: Real-World Impact

Army of Bots isn’t just a fun project — it’s now critical infrastructure for my daily workflow. I use it to:

• Manage operations at Show of Hands retail store (custom bots trained for product uploads, customer service templates, scheduling, and more).
• Write songs for my AI music artists (each artist has their own bot with unique personalities, tone of voice, and lyrical styles).
• Format massive batches of text, code snippets, and content for my various websites and games.
• Brainstorm ideas, troubleshoot problems, and even generate new AI-based workflows on demand.

Thanks to the modular design, adding a new bot now takes less than 10 minutes from idea to deployment — including voice integration and a fully styled avatar.

Bot Creation Pipeline

I even built a custom bot inside Army of Bots — Luna Bot Builder — who writes new systemContext and assistantContext blocks for me. Creating a new bot now involves:

• Using Luna to generate high-quality context prompts.
• Duplicating the PHP + JS template files.
• Assigning the bot to the proper API key.
• Generating a new avatar in Midjourney.
• Plugging it all in — and the new bot is ready to use within minutes.

This pipeline makes expanding my army of assistants effortless and scalable.

Challenges and Lessons Learned

• How to design scalable modular architecture for AI systems and voice integration.
• How to work with two separate AI APIs (OpenAI + ElevenLabs) while maintaining clean UX and minimal latency.
• The power of image analysis to supercharge bot capabilities beyond basic chat.
• How modular thinking can turn a "small" project into a real operational backbone for daily life.

Next Steps

• Expand the bot roster even further to include new AI projects, game assistants, and content moderators.
• Introduce simple database storage to allow bots to "remember" conversations when triggered.
• Develop agents and/or "teams" of bots that can cross-collaborate inside the same interface for multi-perspective brainstorming and problem-solving.