Menu
Mattering Automations connector

A Voyage AI MCP for retrieval that actually retrieves

Connect one API key and Voyage's retrieval models become workflow steps: text embeddings, document reranking, multimodal embeddings over text and images, and contextualized chunk embeddings built for RAG. Four nodes, because that is the entire API, and every one of them is safe.

4 ready-made nodes One credential to connect Works with Claude, ChatGPT, and Matter

What the Voyage AI MCP connector does

The Voyage AI connector for Mattering Automations is 4 nodes, and that is the whole model surface on purpose: Voyage AI: Create Text Embeddings turns text into vectors, Voyage AI: Rerank Documents orders candidate documents by relevance to a query, Voyage AI: Create Multimodal Embeddings embeds inputs that mix text with image or video URLs into one vector space, and Voyage AI: Create Contextualized Chunk Embeddings embeds document chunks so each vector carries its surrounding document context, which is the failure mode plain chunk embeddings have in RAG pipelines.

Every node speaks MCP, the open protocol for connecting AI assistants to tools, so Claude, ChatGPT, and any MCP-capable assistant can use them, alongside Matter, the AI assistant built into the platform. An assistant can embed a query, rerank a shortlist of documents, and hand back only the most relevant ones, through the same nodes your workflows use.

Because Mattering is a work platform rather than middleware, retrieval work lands somewhere useful by default: records in a Mattering database can carry embeddings updated whenever their content changes, and a rerank step can decide which knowledge base article or contract clause a workflow surfaces to a person.

There is nothing to be careful about here beyond ordinary API spend. All four nodes are pure compute: they read your inputs, return vectors or rankings, and store nothing on Voyage's side. Voyage meters by tokens processed, and every response reports its token usage so workflows can track cost per run.

One connection, three surfaces

  • Visual automations. Drag Voyage AI nodes onto the canvas and wire them to triggers, conditions, and other apps.
  • AI agents, yours or ours. These nodes speak MCP, the open protocol, so Claude, ChatGPT, and any MCP-capable assistant can use them, and Matter, the agent built into the platform, uses them natively.
  • The whole platform. Combine with native tasks, projects, and other connectors, no middleware in between.

Every Voyage AI node, ready to drop on the canvas

4 nodes, generated from the connector itself so this list always matches what ships. Expand a category to browse.

Action nodes 4

Voyage AI: Create Contextualized Chunk Embeddings

Embed all the chunks of a document together so each chunk's vector carries its surrounding document context. Fixes the RAG blind spot where isolated chunk embeddings lose what their references mean.

Voyage AI: Create Multimodal Embeddings

Embed inputs that interleave text with image or video URLs into one vector space, so a text query can retrieve a diagram or product photo. Media referenced by URL; raw bytes never pass through the workflow.

Voyage AI: Create Text Embeddings

Turn text into vectors, with input type hints (query versus document), output dimension and quantization options. The connector's auth probe: the connection test embeds one short word on Voyage's lightest model.

Voyage AI: Rerank Documents

Order candidate documents by relevance to a query, with scores. The quality step between a wide vector-search shortlist and the documents an LLM or person actually sees.

Connecting Voyage AI takes one credential

Voyage AI issues API keys from the Voyage dashboard (dashboard.voyageai.com). Mattering sends the key as a bearer credential on every request. New accounts include free trial tokens, after which usage is billed per million tokens by model.

1

Create an API key in the Voyage AI dashboard.

2

In Mattering Automations, open Connectors, click Add Connector, choose Voyage AI, and create a credential with that key.

3

Click Test Connection. Voyage has no free endpoint, so the test embeds a single short word on Voyage's lightest model: a real authentication check whose cost rounds to nothing.

4

All 4 Voyage AI nodes appear in the canvas palette.

Common uses for the Voyage AI MCP

Embeddings for semantic search

Voyage AI: Create Text Embeddings turns queries and documents into vectors in one call, with input type hints (query versus document) that Voyage's retrieval models use to embed each side appropriately. Output dimension and quantization options are plain config fields for teams tuning storage against accuracy.

Reranking a shortlist before a model sees it

Voyage AI: Rerank Documents takes a query and candidate documents and returns them ordered by relevance with scores. The standard pattern: cheap vector search produces a wide shortlist, rerank tightens it, and only the top documents reach the LLM. It is the highest-leverage single step in most RAG stacks.

One vector space for text and images

Voyage AI: Create Multimodal Embeddings embeds inputs that interleave text with image or video URLs into a single vector space, so a text query can retrieve a diagram, a product photo, or a slide. Media is referenced by URL, which keeps node payloads light.

RAG chunks that remember their document

Voyage AI: Create Contextualized Chunk Embeddings embeds all the chunks of a document together, so each chunk's vector reflects what the rest of the document says. Chunks like "the fee is 2 percent" become retrievable by what they actually mean rather than only by their literal words.

Example: A knowledge base that answers with its best article, not its first match

Built on the Automations canvas in a few minutes, or described in plain language to Matter and built for you.

1

Trigger: a question arrives from a form or a task in Mattering Productivity.

2

Voyage AI: Create Text Embeddings embeds the question as a query.

3

A Mattering database of knowledge articles supplies the candidate set for the question.

4

Voyage AI: Rerank Documents orders the candidates by relevance and returns scores.

5

The top article and its score land back on the task as a comment, and a low top score routes the question to a person instead of an answer.

Voyage AI MCP questions, answered

Only four nodes?

Yes, and that is coverage, not a subset: text embeddings, reranking, multimodal embeddings, and contextualized chunk embeddings are Voyage's entire model API. The only surface deliberately left out is the Files and Batch API for very large offline jobs, which works by uploading .jsonl files and does not fit a canvas node shape; everything a workflow does interactively is here.

Does this work with Claude, ChatGPT, or other AI assistants?

Yes. Mattering exposes these nodes over MCP, an open protocol for connecting AI assistants to tools, so any MCP-capable assistant, including Claude and ChatGPT, can connect to your workspace and use them. Matter, the assistant built into the platform, uses the same nodes natively.

Is anything in this connector risky to automate?

No. All four nodes are pure compute: they take your inputs, return vectors or rankings, and create or delete nothing on either side. The only thing to manage is API spend, which Voyage meters by tokens processed; every response reports its token usage, so a workflow can log cost per run.

What does it cost?

Voyage bills per million tokens processed, with rates varying by model, and new accounts include free trial tokens; the connector never invents prices, so check the Voyage pricing page for current numbers. Retrieval workloads are typically cheap: embeddings and reranks measure in fractions of a cent, which is also why the connection test, a single-word embedding on the lightest model, costs effectively nothing.

What are contextualized chunk embeddings and when do they matter?

A fix for RAG's classic blind spot. Plain chunk embeddings encode each chunk in isolation, so a chunk saying "this fee applies after the first year" loses what "this fee" means. Voyage AI: Create Contextualized Chunk Embeddings embeds a document's chunks together so each vector carries surrounding context. They matter most for contracts, policies, and documentation, where chunks constantly refer to things defined elsewhere in the document.

How do the multimodal embeddings handle images and video?

By URL. Inputs interleave text segments with publicly reachable image or video URLs, and Voyage embeds the combination into the same vector space as text-only inputs, so one index serves mixed content. Raw media bytes never pass through the workflow, which keeps node outputs small.

Which models are available through the nodes?

The model is a plain config field on every node, so new Voyage models work the day they ship, without a connector update. Voyage's lineup spans general-purpose and lightweight embedding tiers, domain-tuned variants for code, finance, and law, rerankers, and the multimodal and contextualized models; the Voyage docs list current model names and their strengths.

4 nodes waiting on the canvas

Put Voyage AI to work inside your workspace

Connect once, then use every node in visual automations and AI agent workflows across the Mattering platform.

All product names, logos, and brands are property of their respective owners and are used for identification purposes only. Use of these names, logos, and brands does not imply endorsement or affiliation. Voyage AI is a product of its respective owner; Mattering is an independent platform.

Start Free Explore our Products