Unbody is a new kind of backend built around knowledge instead of traditional data storage.

Instead of rows, tables, and rigid schemas, Unbody revolves around rich, semantic knowledge that your backend can actively understand, reason about, and act upon at runtime.

Knowledge-base vs. Database

Traditional backend stacks store data as isolated, static pieces of information—often structured into tables and rows with fixed schemas.

This works great for applications that retrieve and store fixed, predictable information—but modern AI-native applications have fundamentally different needs.

• Users don’t just want to fetch information—they ask questions.
• Applications don’t just store facts—they must understand meaning, context, and intent.
• Software no longer just serves static responses—it dynamically reasons, summarizes, generates content, and acts autonomously.

To power these capabilities, you need more than a database. You need a knowledge-base.

A knowledge-base is a structured, semantic graph of information—meaningfully enriched with context, relationships, embeddings, and metadata. It can handle multimodal content seamlessly—text, images, videos, audio—and makes them semantically searchable, contextually retrievable, and runtime actionable.

How Unbody Fabricates Knowledge

Turning raw data into meaningful, semantic knowledge is not trivial. Unbody makes this process seamless through a structured pipeline called knowledge fabrication:

1. Data Ingestion:
   Pull content from APIs, databases, websites, PDFs, videos, or images.

2. Parsing and Structuring:
   Automatically transform diverse formats into structured, consistent representations.

3. Enhancement & Transformation:
   Semantic agents automatically extract context, meaning, entities, relations, and metadata from your data—making it queryable and actionable.

4. Vectorization:
   Structured knowledge is embedded into semantic vectors, enabling semantic search, reranking, and powerful reasoning capabilities.

The result is a dynamic knowledge-base optimized for semantic retrieval, reasoning, and intelligent interactions.

Why This Matters (Philosophy)

We believe software is becoming closer to humans than ever before. We no longer interact by merely clicking buttons or filling forms—we interact through natural language and rich context.

This fundamental shift means your backend must also evolve—it can no longer be passive storage. It must actively understand, reason, and respond intelligently.

Unbody is designed precisely for this world—making it straightforward to build intelligent backends from scratch, without complexity and without compromise.

Core Unbody Principles

• Knowledge-first, not data-first
  Store meaning, relationships, and context—not just isolated records.

• Runtime Intelligence
  Reason, generate content, and dynamically respond at runtime.

• Multimodal by Default
  Handle text, images, videos, and audio seamlessly—just like humans.

• Modular and Flexible
  Swap components easily; use built-in tools or plug in your own.

Next Steps

With these foundational concepts clear, you’re ready to explore how Unbody’s architecture implements them technically, or jump right into building:

• → See Unbody’s Architecture
• → Set up your first Unbody backend