Aryn
Aryn is an AI-powered platform for document parsing, data extraction, and analytics, enabling users to process and query unstructured documents at scale.
Verdict
Common use cases
- Organize contract PDFs into queryable docsets
- Retrieve document metadata for compliance audits
- Generate query plans before running analytics
- Check async task status during bulk uploads
- Delete outdated docsets after project closure
Integration
- Vendor
- Aryn
- Category
- docs
- Auth
- API_KEY
- Tools
- 8
- Composio slug
aryn
Tools
- Create DocSet
Tool to create a new docset. use when you need to allocate a storage container before adding documents.
- Create DocSet
Tool to create a new docset. use when you need to allocate a storage container before adding documents.
- Delete DocSetdestructive
Tool to delete a docset and all its documents. use after confirming the docset id, when you need to permanently remove a docset and its contents.
- Generate plan
Tool to generate a query plan without executing it. use when you need to review the logical plan before running your query (e.g., "generate a plan for revenue breakdown by region").
- Get DocSet Metadata
Tool to retrieve metadata for a specific docset. use when you have the docset id and optionally need usage statistics.
- Get Document Binary
Tool to retrieve the binary content of a document by docset id and document id.
- Get Document by ID
Tool to retrieve a document by id. use after both docset and document ids are available.
- List Async Tasks
Tool to list all outstanding asynchronous tasks for the account. use when you need to check pending or running tasks.
Setup
Setup guide
- 11. In Switchy, open your workspace settings and navigate to the Integrations tab. 2. Click 'Add MCP' and select Aryn from the catalog. 3. Log into your Aryn account at aryn.ai, navigate to API settings, and generate a new API key with read and write permissions. 4. Paste the key into Switchy's auth field and click 'Connect'. 5. Switchy will verify the key and list available tools in the connection summary. 6. Open any Space, type '@Aryn create a docset called Q1Reports', and send the message. 7. Aryn will return a docset ID — save it in the Space's pinned notes so future prompts can reference it. 8. Test retrieval by uploading a document to that docset via the Aryn dashboard, then ask '@Aryn list documents in docset [ID]' to confirm the connection works end-to-end.
What teammates see: by default, memories from Aryn are scoped to the Space (PROJECT visibility) - you can mark any memory PRIVATE or share it ORG-wide.
Works well with
Top models
Compatibility data appears once enough Spaces have used this MCP together with a given model.
How Switchy teams use it
Starter prompts
Create Research Docset
@Aryn create a docset called TeamResearch2024. Once it's ready, give me the docset ID so I can start adding documents.Open in a Space →
Retrieve Document Binary
@Aryn get the binary content for document ID doc_abc123 in docset ds_xyz789. Save it so I can review the original file.Open in a Space →
Generate Query Plan
@Aryn generate a query plan for 'revenue breakdown by region in Q4' against docset ds_finance2024. Show me the plan before running it.Open in a Space →
Check Pending Tasks
@Aryn list all outstanding async tasks for my account. Let me know which ones are still running and which completed.Open in a Space →
Delete Old Docset
@Aryn delete docset ds_archive2023 and confirm once all documents are removed. I've verified this is the correct ID.Open in a Space →
Example outputs
Illustrative - representative of the model's voice and quality, not literal recordings.
@aryn list all docsets in my account and show me which ones have the most documents stored
Based on the metadata retrieved, you have 4 active docsets. Your 'Q4-2024-Reports' docset contains 127 documents (largest), followed by 'Customer-Onboarding' with 89 documents, 'Legal-Contracts' with 34 documents, and 'Product-Specs' with 12 documents. Total storage across all docsets is approximately 2.3 GB. All docsets were last updated within the past week.
This example demonstrates Aryn's read capabilities for inventory management. The MCP retrieves docset metadata to surface storage patterns. Requires API key with read permissions. Response times scale with the number of docsets in your account—expect 1-3 seconds for typical workloads.
@aryn create a new docset called 'Engineering-RFCs' for storing technical design documents, then upload the three RFC files from my desktop
DocSet 'Engineering-RFCs' created successfully with ID ds_8x4k2m9p. Now uploading 3 documents... Upload complete: 'RFC-001-API-Gateway.pdf' (2.1 MB), 'RFC-002-Cache-Strategy.pdf' (1.8 MB), 'RFC-003-Database-Migration.pdf' (3.4 MB) are now indexed and searchable. The docset is ready for querying. Note that full-text indexing may take 2-3 minutes for optimal search performance.
This example shows Aryn's write workflow: creating a storage container and populating it with documents. The MCP handles both the docset allocation and asynchronous document ingestion. Be aware that large uploads (>10 MB per file) may trigger rate limits, and indexing lag means immediate queries might miss newly added content.
@aryn generate a query plan to find all documents in 'Q4-2024-Reports' that mention revenue targets above $5M, then summarize the key findings
Query plan generated: The system will (1) scan 127 documents in docset 'Q4-2024-Reports', (2) filter for mentions of revenue figures >$5M using semantic search, (3) rank results by relevance. Estimated 8-12 matching documents. Executing plan... Found 9 relevant documents. Summary: Your Q4 reports show three product lines exceeded $5M targets—Enterprise SaaS ($7.2M, +18% YoY), API Platform ($6.1M, +12%), and Analytics Suite ($5.4M, +9%). Two regions (EMEA, APAC) are tracking behind forecast but show Q4 acceleration.
This example pairs Aryn's query planning tool with AI synthesis to extract insights from document collections. The 'generate plan' step lets you validate search logic before execution—useful for complex queries over large docsets. Semantic search quality depends on document structure; PDFs with clean text extraction perform better than scanned images.
Use-case deep-dives
When Aryn beats manual doc search for support teams
A 6-person support team fields 40 tickets a day, half of which need answers buried in product docs, internal runbooks, or past case notes. Aryn wins here because the query-plan tool lets agents preview how the search will run before executing—useful when the question spans multiple doc sets (product specs, troubleshooting guides, compliance PDFs). The docset structure keeps customer-facing docs separate from internal notes, so agents don't accidentally leak internal context. The binary-retrieval tool pulls original PDFs when a ticket needs an exact screenshot or diagram. If your knowledge base is under 10k documents and you already have API-key infrastructure, Aryn slots in fast. Beyond 10k docs or if you need real-time collaboration on answers, you'll want a heavier RAG stack.
Aryn for versioned onboarding materials in fast-growth teams
A 12-person startup onboards 3 new hires a month, each needing role-specific docs (eng setup guides, sales playbooks, design system PDFs). Aryn's docset model maps cleanly to this: one docset per role, updated async as the team scales. The async-task tool tracks when a new batch of onboarding PDFs finishes indexing, so the hiring manager knows when to send the link. The metadata tool shows which docsets are stale (last updated 6 months ago), flagging when it's time to refresh the sales deck. This works until you hit 50+ hires a quarter—at that scale, you need a full LMS with permissioning, not a doc-retrieval MCP. For startups under 30 people, Aryn keeps onboarding docs queryable without Notion bloat.
When Aryn handles contract review for small legal ops
A 3-person legal-ops team reviews 20 vendor contracts a month, extracting liability caps, termination clauses, and data-residency terms. Aryn's binary-retrieval tool pulls the original signed PDF when a clause needs visual confirmation (tables, redlines, signatures). The query-plan preview helps the paralegal verify the search will hit all relevant sections before running a bulk extraction across 50 contracts. Docsets separate active contracts from archived ones, so queries don't surface expired terms. This setup works for teams processing under 100 contracts a month with straightforward clause types. If you need multi-party negotiation tracking or clause-change diffs, you're past MCP territory and into contract-lifecycle platforms. For small legal ops doing post-signature retrieval, Aryn is the right weight.
Frequently asked
What does the Aryn MCP do in Switchy?
It lets your team create and manage document storage containers (DocSets), upload files, retrieve document metadata, and generate query plans against those documents. Think of it as a programmable document repository you can query through natural language in Switchy, without building your own indexing pipeline.
Do I need special permissions to connect Aryn?
You need an Aryn API key with write access to create and delete DocSets. If you only plan to read existing documents or generate query plans, a read-only key works. The MCP uses API key authentication, so whoever connects it controls what the team can do in Aryn through Switchy.
Can the Aryn MCP search inside document content?
Not directly. The MCP retrieves document binaries and metadata by ID, and generates query plans, but it doesn't execute full-text search or semantic retrieval on its own. You'd use the query plan tools to structure searches, then fetch results by document ID. For live search, query Aryn's API separately.
How is this different from just using Aryn's dashboard?
The dashboard is for manual setup and monitoring. The MCP lets your team automate document ingestion, deletion, and query planning from inside Switchy conversations. If you're building workflows that need to add documents or check task status on the fly, the MCP saves you from context-switching to the Aryn UI.
Who on the team should connect this integration?
Whoever manages your document pipelines or has access to your Aryn account. They'll need the API key and should understand which DocSets exist. Once connected, anyone in the Switchy workspace can trigger Aryn tools, so limit access if you have sensitive document sets.