VATES is an AI B2B knowledge infrastructure designed for organizations that need reliable answers from information they control. VATES can be used as a standalone conversational interface, embedded into existing websites, or integrated into AI products and processes via API.

The core differentiator lies in how VATES retains and utilizes knowledge. Instead of turning organizational knowledge into embedding vectors, VATES keeps knowledge in the form of structured text that humans can read and edit directly. Knowledge is interpreted at the moment a question is asked.

Main features:

1. No use of vector databases or embedding pipelines. No need for Pinecone, Weaviate, or pgvector, and no GPU is required for the knowledge retrieval system.

2. Direct updates. When knowledge content is edited or deleted, changes take effect immediately without needing to recreate embeddings or index vectors.

3. One knowledge base, multiple languages. The knowledge base can be written in one language while users ask questions and receive answers in another. No need to maintain separate translations of the same knowledge base.

4. Readable and traceable knowledge. Content is stored as structured text rather than numerical vectors, allowing humans to directly inspect the stored knowledge and trace the basis of the answers.

5. Multiple integration methods. VATES operates as a standalone interface via URL, an embedded widget with a single line of JavaScript, API, and remote MCP server with OAuth 2.1.

Support for Vietnamese: The introductory page, administration dashboard, and operational guidelines of VATES all have a Vietnamese version. VATES supports the interface and documentation in 23 languages across 9 writing systems. The language of the knowledge base and the conversational language do not need to match; for example, a knowledge base written in English can still be used to answer questions in Vietnamese.

Use cases: VATES is suitable for organizations with proprietary information that frequently changes, where the information owner decides the correct answers and incorrect answers have real costs. Examples include customer support based on internal policies, operational knowledge bases, and providing product or service information that is regularly updated.

How to get started: VATES is a web-based platform and can be used immediately via a standalone URL, embedded into a website using a line of JavaScript, or integrated via API. The administration dashboard also works on mobile devices. Payment is based on usage with a prepaid balance; the minimum initial deposit is $5, with no monthly subscription fees or cancellation charges.

Design platform: The architecture of VATES has been developed alongside independent research on knowledge representation, observer-dependent interpretation, and communication between heterogeneous intelligences. VATES is not designed as an additional optimization layer for vector-based RAG pipelines; it uses a different approach, where knowledge can still be read and edited directly until the point of interpretation.