v2v: From Documents to an Interactive Knowledge Graph

Note: All claims are based on the author's own description. No revenue, customers, or independent verification is available.

The description states that v2v is a DAG-based AI workflow platform built in Go. It performs:

• Text extraction from PDFs, HWP, DOC/DOCX, XLS/XLSX, emails, and images
• OCR processing for embedded images
• Configurable text chunking strategies
• Storage and retrieval using vector databases (e.g., Qdrant)
• Workflow reuse through "Workflow Reference Vertices"
• Generation of interactive radial knowledge graphs as static HTML reports
• Dynamic topic input and discovery from retrieved evidence

It uses technologies including:

• Go for core engine and Vertex implementations
• Qdrant, pgvector, Weaviate, Milvus, OpenSearch, Elasticsearch, Chroma, Redis Vector, Vespa
• Docker for local vector DB environments
• Codex with the 5.6 SOL model during development
• OCR and document parsing libraries

Inference: The system is described as modular and composable via a DAG structure where each step is a "Vertex", suggesting a visual workflow engine.

The author positions v2v as a tool that turns unstructured data (documents, messages, web pages) into reusable AI workflows and interactive knowledge graphs. It aims to avoid creating large, hard-coded pipelines for every use case by enabling workflow composition.

Key claims:

• Enables users to enter a topic and generate a new knowledge report without predefined categories.
• Reuses completed workflows through “Workflow Reference Vertices”.
• Generates an interactive radial knowledge graph with detailed node exploration.
• Supports multiple input formats and vector databases.
• Designed for local, cloud, and enterprise environments.

Claim vs Fact: These are self-reported claims about functionality and design intent. No evidence of actual usage or adoption is provided.

The description does not clearly define a target customer or ideal customer profile (ICP). It implies the platform could be used by organizations with knowledge stored in various document formats, but no specific persona or industry is named.

Not evidenced: No indication of who uses this system, what their job titles are, or how they would interact with it beyond the hackathon demo.

There is no evidence of a business model or pricing structure in the description. The project is presented as a hackathon submission and not as a commercial offering.

Not evidenced: No mention of monetization, licensing, subscriptions, or any revenue-generating mechanism.

The system is built using:

• Go for workflow engine and Vertex implementations
• Qdrant and other vector databases (pgvector, Weaviate, etc.)
• Docker for local environments
• OCR and document parsing libraries
• Optional LLM-based analysis
• HTML/CSS/SVG/JS for interactive reports

It supports:

• Multiple input formats: PDF, HWP, DOC/DOCX, XLS/XLSX, email, image
• Runtime topic input
• Workflow composition via “Workflow Reference Vertices”
• Dynamic graph generation from retrieved evidence

Inference: The modular architecture suggests scalability and extensibility, but no production deployment or performance data is shared.

The project is described as a hackathon submission for the OpenAI 2026 hackathon. It includes:

• A working end-to-end demonstration
• Full workflow from input to output (HTML report)
• Reusable components and workflow composition

However, there is no evidence of:

• Revenue or monetization
• Customers or user base
• Product-market fit or market traction
• Post-hackathon development or iteration

Absence of evidence: No signs of real-world usage or product maturity beyond the demo.

The description does not reference competitors directly. However, it implies a space involving:

• Document processing and extraction tools
• Knowledge graph generation platforms
• Vector databases for semantic search
• Workflow automation systems (e.g., DAG-based engines)

It overlaps with areas like RAG (Retrieval-Augmented Generation), document AI, and knowledge management.

Not evidenced: No competitive analysis or differentiation from existing tools is provided.

• Unproven commercial viability: The project is a hackathon submission with no evidence of traction or monetization.
• No clear business model: No indication of how the platform would be sold or funded.
• Single-founder team: Only one member listed, which may limit execution capacity.
• Self-reported only: All features and claims are unverified by third parties.
• Limited scope in demo: The end-to-end workflow works for one topic but lacks broader testing or validation.

Inference: If this were to evolve into a product, it would face challenges around scalability, data privacy, and integration complexity.

1. What is the actual use case or problem you're solving in real-world settings?
2. How do you plan to monetize this platform beyond the hackathon?
3. Have you tested the system with real users or organizations?
4. What are the technical limitations of running this at scale?
5. How does v2v handle data privacy and security, especially when dealing with sensitive documents?
6. Are there any plans for cloud deployment or enterprise integration?
7. What is your roadmap beyond the current demo?