AI Vector Store

ai.vector-store AI v0.1.0

Self-contained semantic memory: embeds text internally and inserts, queries (cosine similarity), or clears an in-memory vector store keyed by name. No external database required — the RAG retrieval layer that pairs with the Embeddings node. A list on the wire is N records: insert embeds ONE VECTOR PER RECORD, reading the text field from each record rather than from the array itself, and reading the metadata field per record too — one batch-level metadata object on every vector quietly ruins metadata-filtered retrieval. A batch answers with { inserted, store, ids }, a single record with { inserted, store, id }. Query and clear address the store once per execution — a query handed a list embeds the FIRST record's text.

The AI Vector Store step on the Studio canvas — The **AI Vector Store** step as it appears on the Studio canvas — input pins on the left, output ports on the right.

Finding it in the library

Search the builder's node library for AI Vector Store (it lives under AI). A single click opens the in-editor docs panel shown here — description, ports, and every property, without leaving the canvas. Double-click (or drag) to add it to the workflow.

AI Vector Store in the node library, with the in-editor docs panel open — The library entry and the in-editor docs panel for **AI Vector Store** — the same reference this page is generated from.

Wired up in the builder

AI Vector Store in a real, runnable flow — captured live from the Studio editor, exactly as it looks on your canvas. The ◈ ports carry the AI Model provider you wire in. This is the same workflow used for the example input & output below.

AI Vector Store wired into a runnable workflow in the Studio builder — **AI Vector Store** wired into a runnable flow — input on the left, output on the right, AI Model provider wired into the ◈ ports below.

How it’s configured

The node’s settings as the builder shows them — every field laid out with real values. In the Studio these are edited on the node: click the chevron on the divider under its ports to open them.

The AI Vector Store node's settings in the Studio builder — The settings for **AI Vector Store**, showing the values from the flow above.

Ports

Ports are the node’s contract with its neighbours. In the editor a port label renders bold when wired and italic when optional; ◈ ports accept attachment carriers rather than data wires.

Direction	Port	Label	What flows through it
Input	`input`	Input
Output	`output`	Result

How data flows through it

AI Vector Store consumes the content of the incoming envelope — when it is fed directly by a trigger, the trigger’s wrapper is unwrapped at the node boundary so the node sees the actual data, not the metadata shell. Its output becomes the payload for the next node, while the envelope (trace ids, correlation, binary refs) rides along untouched. In the Runs view you always see the whole envelope for both sides of this node.

Expressions in the config

String-typed properties accept {{ }} expressions evaluated against the incoming item at run time — e.g. {{ $json.customer.email }}. On this node that’s store, keywordWeight, backendUrl, backendKey. JSON- and code-typed fields never interpolate — they are passed through literally.

Build it with AI

Every node in this reference is reachable through Flowdrome’s AI Copilot and the MCP tools — say what you want, and the graph surgery happens server-side. Node types resolve fuzzily, so the catalog label (AI Vector Store) works as well as the exact type id (ai.vector-store).

In the Copilot panel (or any connected AI):

add a ai vector store node after the trigger

As a step in a create_chain_workflow call:

{"type":"AI Vector Store","config":{}}

Raw MCP call — add this node to a workflow with add_node

curl -s -X POST http://localhost:4800/mcp -H "content-type: application/json" -d '{ "jsonrpc": "2.0", "id": "1", "method": "tools/call", "params": { "name": "add_node", "arguments": { "workflowId": "<id>", "type": "AI Vector Store" } } }'

Example input & output

Captured from a real test run of the workflow above — this is what the Runs view shows after pressing Test workflow.

Input — what the node received

The AI Vector Store node's input envelope in the run data viewer — The input envelope in the Runs view — Flowdrome always shows the whole envelope, with the payload inside `body`.

Output — what the node produced

The AI Vector Store node's output envelope in the run data viewer — The output envelope after the step ran.

Property reference

Every setting, with its type and default — the same fields shown configured above.

Property	Type	Default	Description
Operation `operation`	`select`	`"query"`	insert = embed the text and add it to the store; query = find the nearest stored texts by cosine similarity; clear = empty the store. `clearinsertquery`
Store name `store`	`string`	`"default"`	Names the in-memory vector store. Different names are isolated collections.
Text field `textField`	`field`	`""`	Dot-path to the text to store (insert) or the query text (query). A LIST on the input is N RECORDS, so the path is read from EACH record — ten items with "content" store ten vectors (a query embeds the first record's text). Blank = the whole payload; on insert a field holding an array of strings stores each element. Shown when `(operation ?? "query") !== "clear"`
Metadata field `metadataField`	`field`	`""`	Optional dot-path to an object stored alongside each vector and returned with matches. Read from EACH record of a list, so every vector carries ITS OWN metadata (one shared tag on all of them makes metadata-filtered retrieval useless). Shown when `(operation ?? "query") === "insert"`
Top K `topK`	`int`	`4`	How many nearest matches a query returns. Shown when `(operation ?? "query") === "query"`
Keyword weight `keywordWeight`	`string`	`"0"`	Hybrid retrieval: blend exact keyword overlap with vector similarity. 0 = pure vector (default); 0.3 lifts exact-term hits (names, ids, codes) a semantic vector can miss; 1 = pure keyword. Shown when `(operation ?? "query") === "query"`
Backend `backend`	`select`	`"memory"`	memory = the built-in self-contained store (no external DB). http = an external vector service (Pinecone/Qdrant/pgvector adapter, or any store speaking the POST /insert /query /clear contract); embeddings are still computed here and the vectors sent to it. `httpmemory`
Backend URL `backendUrl`	`string`	`""`	Base URL of the external vector service (backend = http). POSTs {URL}/insert, /query, /clear. Shown when `backend === "http"`
Backend key `backendKey`	`string`	`""`	Optional Bearer key for the external service — supports ${credential.NAME.FIELD}. Shown when `backend === "http"`
Timeout (ms) `timeoutMs`	`int`	`120000`	Abort the embedding request after this many milliseconds. Shown when `(operation ?? "query") !== "clear"`

Related nodes

The rest of the AI group — the same folder you’d scan in the editor’s library.

This page is generated from the node registry by gen-node-docs.mjs on every site build — ports, properties, defaults and visibility rules cannot drift from the code. The screenshots and example data are captured from a live Flowdrome by npm run shots:nodes and npm run gen:examples.