Entamy VecSearch — find documents by meaning, not just matching words

Find the right knowledge even when you cannot remember the exact words. Entamy VecSearch searches specifications, reports and goals in the Entamy administration platform by semantic proximity as well as literal text. A short question can lead directly to the decision or evidence buried inside a long document.

Different words, relevant answers#

The wording in a document rarely matches the words a person uses when searching. VecSearch maps text into a space of meaning and ranks documents by their proximity to the question. It reduces the time spent inventing the perfect query for a specification whose name you no longer remember.

Three search modes, one entry point#

• Semantic — find specifications with related meaning even when the wording differs
• Keyword — locate exact product names, API names and error codes
• Hybrid — combine semantic proximity with literal matches for both breadth and precision

Results can be narrowed by product, document kind, and title or body scope. People searching in the administration interface and AI or external services using the API share the same search entry point.

D1 is the source of truth. Vectors are a rebuildable index.#

VecSearch does not move the canonical document just to make search convenient. Titles, bodies and versions remain in D1. Vectorize stores embeddings with document IDs, versions and chunk positions; document bodies are not copied into vector metadata.

The service regularly reconciles additions, updates and deletions, so it can recover even if an update signal is missed. If the index ever needs repair, it can be rebuilt in full from the latest D1 content.

Protect usage limits without stopping search#

Usage for Workers AI and Vectorize is visible in the administration interface. When the configured safety threshold is reached, only scheduled index creation and updates can pause. Human and API searches continue, as do edits to the D1 documents. Indexing can resume automatically after the daily allowance resets.

Turn Entamy knowledge into usable infrastructure#

Stored documentation creates value only when people and systems can find it at the moment of need and return to the source text behind an answer. VecSearch is the retrieval layer that connects questions to decisions across the Entamy Platform.