Botmanager

• 

  BotManager Dashboard & Architecture Overview

Inspiration

Let’s face it—modern workflows demand too much juggling. We’re constantly bouncing between apps, and we wanted to build something that actually brings order to the chaos. Our goal was to create a unified platform that combines smart chat interfaces with real, heavy-lifting backend automation, giving anyone the power to effortlessly manage custom AI chatbots and dynamic MCP (Model Context Protocol) agents all in one place.

What it does

BotManager is a modular AI powerhouse built for creating, managing, and deploying custom chatbots and dynamic MCP agents powered by tailored knowledge bases. It hooks directly into Google Workspace with secure integration to handle email workflows and personal digital tasks—all controlled from a slick, intuitive dashboard.

How we built it

• Backend & API: We used FastAPI paired with Laravel 12 to handle the heavy routing, controllers, and multi-tenant logic.
• AI & Agents: We wired in LangChain, ChatGroq, and Hugging Face inference models to power the core conversational logic and document understanding.
• Knowledge Base & Storage: We set up high-speed vector search and document retrieval using ChromaDB backed by Supabase (PostgreSQL).
• Frontend: The UI is built using clean Blade templates and styled nicely with Tailwind CSS.
• Automation & Security: We integrated secure OAuth through the Google Cloud Platform so users can connect their Google Workspace and Gmail safely.

Challenges we ran into

Getting fine-grained Google OAuth scopes to cooperate and handling token re-authorization for the Gmail API gave us a run for our money. We finally conquered it by tweaking the Google Auth Platform data access settings and enforcing offline access tokens. On top of that, fine-tuning multi-tenant vector storage for custom knowledge bases took some serious database patience.

Accomplishments that we're proud of

• Successfully pulling off a hybrid FastAPI and Laravel architecture that seamlessly blends modern web views with high-speed AI inference.
• Rolling out secure Google Workspace integration that actually makes account connection and automated data handling smooth.
• Building a truly cohesive ecosystem that ties together embeddable widgets, live MCP agents, and vector-backed document retrieval.

What we learned

This project leveled up our skills in multi-tenant app design, wrestling with complex Google Cloud OAuth permission flows, and turning Model Context Protocol concepts into working, real-world AI agent architectures.

What's next for BotManager

• Expanding our MCP agent toolsets to plug into platforms like Slack, Jira, and GitHub.
• Boosting dashboard analytics so users can easily track agent interactions and token consumption.
• Scaling up our cloud deployment setups for enterprise-grade multi-tenant isolation.

Built With

• chromadb
• fastapi
• google
• groq
• langchain
• laravel-12
• python
• supabase
• tailwind-css