About

How our system works.

Welcome to Our Training Platform

Discover how our innovative system empowers you to create, manage, and share AI training datasets tailored for tools like Vanna AI, LangChain, and beyond. Whether you're building private knowledge bases or contributing to open-source intelligence, our platform provides a structured, secure, and collaborative environment to supercharge your AI projects.

Core Structure: Projects, Plans, and Trainings

At the heart of our platform is a hierarchical system designed for scalability and organization. Here's how it works:

  • Projects: The top-level container for your AI training initiatives. A project groups related work and defines access levels (more on this below).
  • Plans: Within each project, create plans to outline specific training objectives or phases. Plans act as blueprints, ensuring your trainings align with project goals.
  • Trainings: The building blocks inside plans. We support four core types of trainings to cover diverse data ingestion needs:
    • DDL (Data Definition Language): Define database schemas, tables, and structures to lay the foundation for your AI's understanding.
    • SQL: Add sample queries and results to teach your AI how to interact with data effectively.
    • Documents: Upload text files, PDFs, or other unstructured content to infuse domain-specific knowledge.
    • Question and Answer: Pair user queries with expert responses to fine-tune conversational and retrieval capabilities.
Pro Tip: Start with Metadata SQL Initialization in your plan. This essential step maps tables, fields, and relationships in your database schema. It serves as the root node, allowing all subsequent trainings (DDL, SQL, etc.) to attach as sub-nodes. This ensures your AI has a clear, relational view of the data from the outset.

Project Visibility and Access: Tailored for Every Need

Control who sees and uses your project with flexible visibility options. Each project type balances privacy, collaboration, and monetization:

Project Type Description Access & Sharing Monetization
Private Ideal for internal or user-specific data (e.g., company secrets or personal datasets). Not listed publicly; downloadable only by the owner and authorized users. Includes service and storage fees for secure hosting.
Public Listed for broad visibility, fostering community-driven AI growth.
  • User/Group-Created: Standard public sharing from individuals or teams.
  • Open Source: Collaborative mode where users can contribute knowledge items (e.g., new Q&A pairs or documents).
Fully downloadable by anyone. 		Flexible: Free access, donations, one-time download fees, or recurring membership for premium features.
Protected Owned by AI trainers; perfect for valuable, proprietary datasets that need controlled distribution. Hidden from public searches; access requires purchase, permission request, or approval. Training data remains shielded from external inputs or views. Mirrors public sales models (donations, fees, memberships) but with added security layers.

Advanced Settings: Security and Quality Control

Go beyond basics with features that ensure trust and compliance:

  • Vetting Process: Enable consumer vetting for sensitive or x-rated trainings. Review applicants before granting access, protecting your data and maintaining community standards.
  • Customization: Integrate with Vanna AI for SQL generation, LangChain for chaining workflows, or other frameworks via exported datasets.
Getting Started: Create your first project today—it's free to begin! Dive into plans, add your metadata schema, and watch your AI trainings come alive.
Start a New Project