VATES - Showcase Project

VATES is embeddable AI infrastructure for B2B products, built and operated by STANDOUT Inc. in Okayama, Japan. Developers integrate it through an API, the way they would integrate Stripe or Twilio. It also runs as a ready-to-use widget (one line of JavaScript), as a standalone chat URL, and as a remote MCP server with OAuth 2.1. What makes it structurally different from RAG: knowledge is held as structured, human-readable plain text and interpreted when a question is asked. There is no vector database, no embedding pipeline, and no re-ranking step. Editing a line takes effect immediately with no re-indexing. Every answer can be traced back to the stored knowledge it came from, so the knowledge base and its sources remain directly inspectable. Because knowledge is never translated or fixed in advance, the language of the knowledge base and the language of the conversation are independent. A knowledge base written in one language can answer a user in another without storing translated copies. The interface, documentation, and public site are localized in 23 languages across 9 writing systems. For AI clients, VATES exposes a remote MCP server with OAuth 2.1. MCP operations are idempotent by design. Billing is prepaid and usage-based, with no subscription, monthly fixed fee, or minimum spend. The minimum top-up is USD 5.00.