File System

The shared operational memory for your team and AI agents — upload, organize, search, and access all your sales content in one place.

What is the File System?

The File System is Consuelo’s centralized content layer where both human reps and AI agents store and access persistent sales content. Think of it as the shared operating environment for your entire revenue team — scripts, recordings, battle cards, agent configurations, and any document your team needs.

Every file you upload is automatically indexed for semantic search, meaning you (and your AI agents) can search inside file contents, not just by file name.

Key Concepts

Files

Files are the core unit of the file system. A file is any persistent content that your team or agents need to access — PDFs, documents, spreadsheets, images, recordings, scripts, or agent configuration files.

Files live in the Files page in the sidebar and are also accessible on individual record pages (People, Companies, Opportunities) through the Files tab.

File Categories

Every file has a category that determines which prebuilt view it appears in:

Category	Description	Example
Text Document	PDFs, Word docs, plain text	Contracts, proposals, meeting notes
Spreadsheet	Excel, CSV files	Financial forecasts, contact lists
Presentation	PowerPoint, slides	Pitch decks, training materials
Image	PNG, JPG, screenshots	Product photos, diagrams
Audio	Audio files	Call recordings (future: auto-transcribed)
Video	Video files	Demo recordings, training videos
Archive	ZIP files, folders	Project bundles, backup data
Recording	Call and meeting recordings	Discovery calls, demo walkthroughs
Script	Sales scripts and templates	Cold call openers, objection handlers
Agent File	AI agent configuration files	Steering files, methodology docs, skill definitions
Folder	Uploaded folder contents	Project directories
Other	Anything else	Miscellaneous files

Collections

Collections are workspace-scoped groupings for the knowledge base. When files are indexed, their content chunks are stored in a collection. You can create multiple collections to organize knowledge by topic, team, or use case.

A default collection is automatically created for each workspace. All auto-indexed files go into this collection unless you specify otherwise.

Knowledge Base

The Knowledge Base is the semantic search layer built on top of the file system. When a file is uploaded, Consuelo automatically:

1. Extracts text from the file (PDF, Word, plain text, Markdown, CSV, HTML)
2. Splits the text into chunks (500 tokens each, with 50-token overlap for context)
3. Generates vector embeddings using OpenAI’s text-embedding-3-small model
4. Stores the chunks and embeddings in a pgvector-powered database
5. Makes the content searchable via semantic similarity

This means you can search for concepts, not just keywords. Searching “how to handle pricing objections” will find relevant content even if those exact words don’t appear in the file.

Prebuilt Views

The Files page comes with five prebuilt views that cannot be deleted:

All Files

Shows every file in your workspace. This is the default view when you click Files in the sidebar.

Recordings

Shows files categorized as Audio or Recording. Use this view to find call recordings, meeting recordings, and audio content.

Agent Files

Shows files categorized as Agent File. These are configuration files that AI agents use — steering documents, methodology frameworks, qualification checklists, and skill definitions. When you connect an external AI agent to Consuelo via the GraphQL API, these are the files it reads to understand how your team sells.

Scripts

Shows files categorized as Script. Sales scripts, email templates, objection handling guides, and call frameworks live here.

Temporary

Shows files that have an expiration date set. Temporary files are automatically cleaned up after their expiration passes. Use this for short-lived content like one-time reports, temporary exports, or time-sensitive documents.

How Files Connect to Records

Files can be attached to any record in Consuelo:

• People — attach contracts, proposals, or notes to a contact
• Companies — attach NDAs, partnership agreements, or company research
• Opportunities — attach quotes, SOWs, or deal-specific documents
• Tasks — attach reference materials or deliverables
• Notes — attach supporting documents to a note
• Dashboards — attach reports or data exports
• Workflows — attach configuration files or templates

When you view a record’s detail page, the Files tab shows all files attached to that specific record.

Semantic Search via GraphQL API

AI agents and integrations can search file contents through the GraphQL API:

query {
  knowledgeSearch(input: {
    query: "how to handle pricing objections"
    limit: 5
    minSimilarity: 0.7
  }) {
    content
    similarity
    collectionName
    fileId
  }
}

This returns the most relevant text chunks from across all indexed files in the workspace, ranked by semantic similarity.

Managing Collections via API

# list all collections
query {
  knowledgeCollections {
    id
    name
    chunkCount
  }
}

# create a new collection
mutation {
  createKnowledgeCollection(input: {
    name: "Sales Playbook"
    description: "Battle cards, scripts, and methodology docs"
  }) {
    id
    name
  }
}

# manually index a file into a specific collection
mutation {
  indexFileInKnowledgeBase(input: {
    fileId: "file-uuid-here"
    collectionId: "collection-uuid-here"
  }) {
    chunkCount
  }
}

Auto-Indexing

Files are automatically indexed when uploaded. Consuelo checks the file type and, if it’s a supported text format (PDF, Word, plain text, Markdown, CSV, HTML), extracts the content and indexes it into the workspace’s default collection.

Supported file types for auto-indexing:

• PDF (.pdf)
• Microsoft Word (.doc, .docx)
• Plain text (.txt)
• Markdown (.md)
• CSV (.csv)
• HTML (.html)

Images, videos, archives, and other binary files are stored but not indexed for text search.

File Fields

The File object (attachment) has these fields:

Field	Type	Description
Name	Text	The file name
File	File	The actual uploaded file
File Category	Select	The category (Text Document, Script, Recording, etc.)
Expires At	Date & Time	Optional expiration date for temporary files
Created By	Actor	Who uploaded the file
Created At	Date & Time	When the file was uploaded

Tips

Organize with Categories

When uploading files, set the correct category so they appear in the right prebuilt view. A cold call script should be categorized as Script, not Text Document.

Use Collections for Teams

Create separate collections for different teams or use cases. Your SDR team might have a “Prospecting Playbook” collection while your AE team has a “Closing Playbook” collection.

Leverage Semantic Search

Don’t just search for file names — search for concepts. The knowledge base understands meaning, so “enterprise pricing strategy” will find relevant content even if the file is titled “Q4 Deal Desk Guidelines.pdf”.

Temporary Files for Exports

When generating one-time reports or exports, set an expiration date. This keeps your file system clean without manual cleanup.

Next Steps

• File Categories — learn about each category and when to use it
• Knowledge Base — deep dive into semantic search and collections
• Auto-Indexing — how automatic file indexing works
• File System FAQ — common questions and answers