Pinecone
Put Pinecone to work on your own records, inside your rules, with a person approving what it changes.
25 ways to read, 10 ways to write, each write switched on by you.
Quality checks
Built and tested with example responses.
- Ready to connect: doneEvery listed action is built in.
- Built and tested with example responses: doneEach action is built from the provider’s published API and tested against example responses, errors, rate limits and paging.
Connected, approved, proven.
The things teams do most with Pinecone, an example flow with a person in the loop, and what Nexra keeps as proof.
Top actions
- Create backupApproval by rule
- Create serverless indexApproval by rule
- List assistantsRead
Example flow
- Pinecone · List assistantsDone
- Pinecone · Create backupNeeds approval
- Nexra · Record the decision and result in the evidence chainProven
Example workflow. Routine runs go through on their own. Only a change your rules flag, such as one over a limit you set, waits for the right person, who sees the exact change and the rule before it reaches Pinecone.
What gets logged
- Whether it ran on its own or who approved it, and under which rule
- A fingerprint of the exact request sent to Pinecone
- Pinecone's reply: what it created or changed
- One entry in the evidence chain, linked to the one before
Try this in the sandbox
Run the whole flow with test data in your sandbox, then connect a Pinecone test account before you go live.
For developers
Technical detail
Upsert, fetch and query vectors and text records in a Pinecone index, and manage indexes, backups, imports, inference and assistants for semantic search and RAG.
Pinecone API docs- Source actions
- 25
- Destination actions
- 10
- Data types
- 17
What you can map
Reads are always available. Writes are off until you switch them on for a connection, and governed writes follow your policies.
Assistants2
- Describe assistantSource
One assistant's status, host, region, instructions and metadata.
- List assistantsSource
Pinecone Assistants in the project with status, region and instructions.
Backups3
- Create backupDestinationNot retried automaticallyApproval before production
Creates a backup of a serverless index. Backup storage is billed by size, so each call is never retried automatically.
- Describe backupSource
One backup's status, source index, size and record counts.
- List backupsSource
Backups of all serverless indexes in the project, or of one index.
Collections2
- Describe collectionSource
One collection's status, size, dimension and vector count.
- List collectionsSource
Static copies of pod-based indexes with size, dimension and vector count.
Embeddings1
- Generate embeddingsDestinationNot retried automaticallyApproval before production
Embeds up to 96 texts with a Pinecone-hosted model (Inference API). Billed per token.
Imports2
- Describe bulk importSource
One bulk import's status, progress, record count and error.
- List bulk importsSource
Bulk imports from object storage into the index, with progress and errors.
Index stats1
- Get index statisticsSource
Vector counts per namespace, dimension, metric and fullness, optionally for vectors matching a filter.
Indexes4
- Create serverless indexDestinationProvider duplicate protectionApproval before production
Creates a serverless (managed) index with on-demand read capacity and the given schema fields. The index initialises asynchronously.
- Describe indexSource
One index's host, status, deployment, read capacity and schema fields.
- List indexesSource
All indexes in the project with host, status, deployment and schema.
- Configure indexDestinationNot retried automaticallyApproval before production
Changes an index's deletion protection, tags or dedicated read capacity (node type, replicas and shards). Capacity changes are billed.
Inference models2
- Describe inference modelSource
One hosted model's dimensions, metrics, limits and parameters.
- List inference modelsSource
Hosted embedding and reranking models with dimensions and parameters.
Namespace aliases2
- Describe namespace aliasSource
The namespace an alias currently points at.
- List namespace aliasesSource
Aliases that point at namespaces, optionally for one target namespace.
Namespaces3
- Create namespaceDestinationProvider duplicate protectionApproval before production
Creates a namespace, optionally declaring which metadata fields are indexed for filtering.
- Get namespaceSource
Record count, size and metadata schema for one namespace.
- List namespacesSource
Namespaces in a serverless index with record counts.
Records2
- Search records with textSource
Searches an index with integrated embedding using query text, with optional filter and hosted reranking. Embedding and reranking are billed.
- Upsert text recordsDestinationNot retried automaticallyApproval before production
Creates or replaces up to 96 text records by _id in an index with integrated embedding; Pinecone embeds the mapped text field. Embedding is billed.
Rerank1
- Rerank documentsDestinationNot retried automaticallyApproval before production
Scores documents against a query with a hosted reranking model (Inference API). Billed per rerank unit.
Restore jobs2
- Describe restore jobSource
One restore job's status and percent complete.
- List restore jobsSource
Jobs restoring backups into new indexes, with progress.
Vector batches1
- Upsert vectors in batchDestinationProvider duplicate protectionApproval before production
Creates or replaces up to 500 vectors by ID in one request.
Vector IDs1
- List vector IDsSource
Vector IDs in a namespace, optionally by ID prefix (serverless indexes).
Vector set1
- Fetch vectors by IDSource
Fetches up to 100 vectors by ID in one request; missing IDs are omitted.
Vectors5
- Fetch vectorSource
Fetches one vector with its values and metadata by ID.
- Fetch vectors by metadataSource
Pages through vectors whose metadata matches a filter.
- Query vectorsSource
Finds the most similar vectors to a query vector or an existing vector ID, with an optional metadata filter.
- Update vectorDestinationProvider duplicate protectionApproval before production
Updates one vector's values or merges metadata fields by ID; unspecified metadata is kept.
- Upsert vectorDestinationProvider duplicate protectionApproval before production
Creates or replaces one vector by ID with values and metadata.
Access and limits
- Authentication
- Pinecone API key
- Plans and access
- Starter (free), Builder, Standard and Enterprise projects. Vector ID listing, namespaces, backups and imports are supported only on serverless indexes. Backups, imports and dedicated read capacity need a paid plan.
- Test environment
- The Starter plan can be used for testing (serverless indexes in AWS us-east-1).
- Rate limits
- Per-index read and write unit limits and request size limits (upsert up to 1,000 records or 2 MB); per-project control-plane and inference limits. HTTP 429 when exceeded.
- Pagination
- paginationToken cursors for vector IDs, namespaces, aliases, imports, backups and restore jobs; pagination_token for assistants; queries return top K matches.
- Provider events
- Pinecone does not deliver data webhooks. Flows run on demand, on a schedule, or when a scheduled check finds new or changed records.
Before you build a flow
- Each connection targets one index host for data operations; control-plane operations (indexes, collections, backups, restore jobs, inference and assistants) use api.pinecone.io with the same project key.
- Deletes (indexes, namespaces, namespace aliases, vectors, records, documents, backups, collections, assistants), import cancellation and restoring backups into new indexes are intentionally not implemented.
- Index creation is limited to serverless indexes with on-demand read capacity; pod-based, BYOC, CMEK, integrated-embedding (create-for-model) and collection-sourced indexes are not created. Configure index (deletion protection, tags, read capacity) is billable and never retried.
- Generate embeddings, rerank, text record upsert, text record search and backup creation are billed by Pinecone and are never retried automatically.
- Starting bulk imports (which need object storage integrations), backup schedules, namespace alias writes, the documents API, metadata updates by filter and assistant chat, files and evaluation are not exposed.
- Collections apply to pod-based indexes only.
- API version 2026-07 is pinned.
Developer reference
- Create the Pinecone credentials described above with the narrowest permissions your flows need.
- In Nexra, open Connect, then Connections, then New connection, choose Pinecone, enter its settings and credentials, and tick only the write actions you need. Connection guide
- Test the connection, then use its actions as flow sources and destinations. Flow guide
Machine-readable detail: GET /api/public/connect/connectors/pinecone and the Connections API.