Team Name
DDoxers
Hackathon Track
No Code/Low Code
ps-number
3
Email Address
avanishkasar57@gmail.com
Email Addresses of Team Members
mihiramin2004@gmail.com,dishalalwani.0508@Gmail.com,kundarvinayak2004@gmail.com
Project Description
During the hackathon, we built a working prototype of a Visual AI Workflow Builder for customer support and operations.
It allows businesses to create workflows visually using drag and drop nodes such as:
- Ingest
- Triage
- Policy Check
- Logistics
- Resolve
- Critical Review
- Human Override
- Send Reply
What it does
The platform helps automate real business workflows like:
- customer support handling
- refund / return checks
- FAQ resolution
- order management
- payment generation
- human escalation
What makes it different
Unlike a normal chatbot or simple automation tool, this project gives businesses a clear visual system to design, understand, and control AI-powered workflows.
Why it matters
This project turns messy, manual business operations into:
structured, intelligent, and scalable workflows
That is the core problem we aimed to solve during the hackathon.
We built an AI native operations layer that helps businesses visually orchestrate customer workflows instead of managing them manually across fragmented tools.
Inspiration behind the Project
The inspiration came from a simple but painful business reality:
customer support, approvals, payments, logistics, and AI responses are still handled across too many disconnected tools.
Businesses often rely on:
- WhatsApp chats
- spreadsheets
- manual follow ups
- scattered automations
- overloaded support teams
We wanted to solve this by building a platform where businesses can visually design how work should happen, instead of stitching together operations manually.
What inspired us most was the gap between:
- Powerful AI tools and
- Actually usable business systems.
So we built something that makes AI feel operational, transparent, and controllable - not just intelligent.
Tech Stack
We built this project as a visual AI workflow orchestration platform that combines frontend workflow design, AI decision making, and real automation execution .
Core Stack
Next.js / React – frontend application and dashboard
Tailwind CSS + shadcn/ui – modern, production style UI
React Flow – drag-and-drop workflow builder
Supabase + PostgreSQL – workflow, execution, and business data storage
Firebase Auth – secure authentication
n8n – workflow execution engine
Google Gemini (LLM) – AI reasoning and response generation
Project Repo
https://github.com/vinayak1497/spectrum_rookiesv
Demo Video
https://youtu.be/6uWFdrO14_w?si=tQ9XoMyEnSQkMKdg
Presentation Link
https://docs.google.com/presentation/d/1Nn2r5m7sran_dSYMunHLYuOd7tlLAdqmt8YvXV4RLQg/edit?usp=sharing
Anything Else?
Walk into any small business in India right now Pune, Surat, Nashik. Someone is sitting with three phones, copy pasting the same 15 replies across WhatsApp, email, and Instagram DMs. Every day. 63 million MSMEs run on this person. The tools to fix this exist LLMs, automation, semantic search but every single one assumes a developer is in the room. That gap, between a business owner's domain knowledge and executable automation, has never been closed at the right abstraction level. That's exactly what we built for.
We built a visual canvas where a non-technical operator drags four nodes, describes their business in plain language, and goes live in under 90 seconds. The AI reads every ticket, classifies urgency, retrieves answers from the business's own knowledge base via RAG, and drafts a reply in the customer's language, in the business's voice. Nothing reaches a customer without human approval. That's not a constraint, that's the product's identity. And the system compounds: every approved reply strengthens the knowledge base, so draft approval rates climb from 40% in week one to 80% by week eight not because the model changed, but because the business's own decisions trained it. The same canvas that handles support today handles HR, vendor approvals, and procurement tomorrow. We didn't build a chatbot. We built an operations layer for the businesses that can't afford one.
Rules and Code of Conduct
Team Name
DDoxers
Hackathon Track
No Code/Low Code
ps-number
3
Email Address
avanishkasar57@gmail.com
Email Addresses of Team Members
mihiramin2004@gmail.com,dishalalwani.0508@Gmail.com,kundarvinayak2004@gmail.com
Project Description
During the hackathon, we built a working prototype of a Visual AI Workflow Builder for customer support and operations.
It allows businesses to create workflows visually using drag and drop nodes such as:
What it does
The platform helps automate real business workflows like:
What makes it different
Unlike a normal chatbot or simple automation tool, this project gives businesses a clear visual system to design, understand, and control AI-powered workflows.
Why it matters
This project turns messy, manual business operations into:
That is the core problem we aimed to solve during the hackathon.
Inspiration behind the Project
The inspiration came from a simple but painful business reality:
Businesses often rely on:
We wanted to solve this by building a platform where businesses can visually design how work should happen, instead of stitching together operations manually.
What inspired us most was the gap between:
So we built something that makes AI feel operational, transparent, and controllable - not just intelligent.
Tech Stack
We built this project as a visual AI workflow orchestration platform that combines frontend workflow design, AI decision making, and real automation execution .
Core Stack
Next.js / React – frontend application and dashboard
Tailwind CSS + shadcn/ui – modern, production style UI
React Flow – drag-and-drop workflow builder
Supabase + PostgreSQL – workflow, execution, and business data storage
Firebase Auth – secure authentication
n8n – workflow execution engine
Google Gemini (LLM) – AI reasoning and response generation
Project Repo
https://github.com/vinayak1497/spectrum_rookiesv
Demo Video
https://youtu.be/6uWFdrO14_w?si=tQ9XoMyEnSQkMKdg
Presentation Link
https://docs.google.com/presentation/d/1Nn2r5m7sran_dSYMunHLYuOd7tlLAdqmt8YvXV4RLQg/edit?usp=sharing
Anything Else?
Walk into any small business in India right now Pune, Surat, Nashik. Someone is sitting with three phones, copy pasting the same 15 replies across WhatsApp, email, and Instagram DMs. Every day. 63 million MSMEs run on this person. The tools to fix this exist LLMs, automation, semantic search but every single one assumes a developer is in the room. That gap, between a business owner's domain knowledge and executable automation, has never been closed at the right abstraction level. That's exactly what we built for.
We built a visual canvas where a non-technical operator drags four nodes, describes their business in plain language, and goes live in under 90 seconds. The AI reads every ticket, classifies urgency, retrieves answers from the business's own knowledge base via RAG, and drafts a reply in the customer's language, in the business's voice. Nothing reaches a customer without human approval. That's not a constraint, that's the product's identity. And the system compounds: every approved reply strengthens the knowledge base, so draft approval rates climb from 40% in week one to 80% by week eight not because the model changed, but because the business's own decisions trained it. The same canvas that handles support today handles HR, vendor approvals, and procurement tomorrow. We didn't build a chatbot. We built an operations layer for the businesses that can't afford one.
Rules and Code of Conduct