Unified.to
All articles

How to Build AI Search and Insights Across Google Drive, Slack, Zendesk, and Notion


March 10, 2026

Many B2B SaaS products are adding AI search features that allow their customers to ask questions about their company's internal knowledge and operational data. Instead of manually searching documents, support tickets, or conversations, users can ask natural language questions and receive answers grounded in their organization's data.

For example, a support platform might allow teams to search past Zendesk tickets, Slack discussions, and internal documentation to resolve issues faster. A productivity platform might let employees ask questions about company policies stored in Notion or documents in Google Drive.

The challenge is that enterprise knowledge lives across many systems. Documents are stored in Google Drive, conversations occur in Slack, support issues are logged in Zendesk, and knowledge bases are managed in tools like Notion. Each platform exposes different APIs, schemas, and authentication models.

Unified simplifies this by providing normalized APIs across file storage, messaging, knowledge management, and ticketing platforms. Developers can retrieve documents, conversations, and support records through consistent objects and endpoints, enabling AI search features without maintaining dozens of integrations.

Why SaaS Products Build AI Search Features

AI search is becoming a core capability in many SaaS products.

Examples include:

  • AI knowledge assistants
  • Enterprise search platforms
  • Customer support copilots
  • Internal productivity tools
  • AI-powered help desks

These products allow customers to search across their internal tools using natural language.

Typical capabilities include:

  • Document search across company knowledge bases
  • Conversation retrieval from chat platforms like Slack
  • Support ticket search across help desk systems
  • AI-powered question answering using enterprise data
  • Knowledge discovery across files and internal documentation

By integrating with customer systems, SaaS products can build unified search experiences across multiple sources of knowledge.

An AI search feature typically combines data from several Unified API categories.

CategoryDescriptionKey Objects
File StorageDocuments and files from platforms like Google DriveStorageFile
MessagingConversations from platforms like Slackmessage, channel
Knowledge ManagementKnowledge bases such as Notion or Confluencepage, space, comment
TicketingSupport tickets from platforms like Zendeskticket

Using normalized APIs across these categories allows developers to build search features that span multiple enterprise systems.

Unified File Storage Objects

Unified's [File Storage API](/storage) normalizes files and folders across storage providers.

The primary object is StorageFile, which represents both files and directories.

Key fields include:

FieldDescriptionAI Search Usage
idUnique identifier for the fileInternal reference for indexing
nameFile nameUsed as document title in search
descriptionOptional descriptionAdditional searchable metadata
mime_typeFile formatDetermines how content is processed
sizeFile size in bytesUsed for processing limits
parent_idParent folder IDEnables hierarchical navigation
permissionsAccess control metadataUsed for permission-aware search
created_atFile creation timestampUseful for timeline filtering
updated_atLast modification timestampUsed for incremental indexing
versionProvider revision identifierDetects content changes
download_urlTemporary URL to retrieve file contentsUsed to fetch document text
web_urlProvider link to the fileAllows users to open the original file

Files are retrieved through the storage endpoints:

GET /storage/{connection_id}/file
GET /storage/{connection_id}/file/{id}

To retrieve file content, applications call the file endpoint and then download the content using the download_url field.

Unified Knowledge Management Objects

Unified's Knowledge Management API normalizes knowledge base content across platforms like Notion and Confluence.

The API exposes three primary objects:

ObjectPurpose
pageRepresents a knowledge article or document
spaceRepresents a workspace or knowledge collection
commentRepresents discussions attached to pages

Page Object

Pages contain the primary knowledge content.

Key fields include:

FieldDescription
titlePage title
download_urlURL used to retrieve page content
web_urlLink to the page in the source platform
created_atCreation timestamp
updated_atLast modification timestamp
user_idAuthor or editor
space_idParent knowledge space
typePage type

Pages can be retrieved using:

GET /kms/{connection_id}/page
GET /kms/{connection_id}/page/{id}

Because page bodies are not always returned inline, applications typically retrieve the content through the download_url.

Comment Object

Comments represent discussions attached to knowledge articles.

Important fields include:

  • content
  • content_type
  • created_at
  • updated_at
  • user_id
  • page_id

These fields allow AI search systems to incorporate discussions into knowledge retrieval.

Unified Messaging Objects

Unified's Messaging API normalizes conversations across messaging platforms such as Slack.

Key messaging objects include:

ObjectPurpose
messageIndividual email or chat message
channelMessaging channel or conversation group

Messages contain important search fields such as:

  • message body text
  • sender (author_member)
  • recipients (destination_members, mentioned_members)
  • timestamps (created_at)
  • thread relationships (parent_id)
  • attachments and links

Messages are retrieved using:

GET /messaging/{connection_id}/message
GET /messaging/{connection_id}/channel

Threads can be reconstructed by retrieving messages that share the same parent identifier.

Unified Ticketing Objects

Support systems like Zendesk store customer issues as tickets.

The normalized ticket object includes fields such as:

  • subject
  • description
  • status
  • priority
  • requester
  • assignee
  • created_at
  • updated_at

These fields allow AI systems to retrieve past support cases and provide context when answering questions.

Connecting Customer Integrations

Customers authorize their platforms through Unified Connect.

Typical process:

  1. Your application launches the Unified Connect authorization flow.
  2. The user selects integrations such as Google Drive, Slack, Zendesk, or Notion.
  3. The user authorizes the integration.
  4. Unified redirects back to your application with a connection_id.

Your application stores the mapping:

user_id → connection_id

All subsequent Unified API requests use this identifier to retrieve data for that customer.

Building an AI Search Pipeline

Most AI search systems use a retrieval-augmented generation (RAG) architecture.

A typical implementation includes the following steps.

1. Connect data sources

Customers connect their tools through Unified Connect.

2. Retrieve documents and conversations

Use Unified APIs to retrieve files, knowledge articles, messages, and support tickets.

3. Extract content

Download document bodies using download_url or retrieve message and comment text from API responses.

4. Chunk content

Split large documents or conversations into smaller segments suitable for embedding.

5. Generate embeddings

Use an embedding model to convert text segments into vector representations.

6. Store vectors

Persist embeddings in a vector database such as Pinecone or pgvector.

7. Retrieve relevant context

When a user asks a question, embed the query and retrieve the most relevant segments from the vector database.

8. Generate an AI answer

Pass retrieved context to a language model to generate a response grounded in the customer's data.

This approach allows AI systems to answer questions using enterprise knowledge across multiple systems.

Keeping the Search Index Updated

Enterprise data changes frequently, so search indexes must remain synchronized.

Unified provides webhook events that notify applications when records change.

Examples include:

  • file updates in storage systems
  • new Slack messages or channel changes
  • knowledge base updates
  • new or updated support tickets

Applications can subscribe to these webhook events to trigger re-indexing of updated documents.

The updated_at field can also be used to detect changed content and refresh embeddings.

Supported Platforms

Unified supports a broad ecosystem of integrations across knowledge sources.

Examples include:

File Storage

  • Google Drive
  • Dropbox
  • Box

Messaging

  • Slack
  • Gmail
  • Microsoft Teams

Knowledge Management

  • Notion
  • Confluence
  • Guru

Ticketing

  • Zendesk
  • Jira Service Management
  • Freshdesk

This allows SaaS products to support AI search across a wide range of enterprise tools.

Why Product Teams Choose Unified

Building AI search across enterprise tools normally requires integrating with many APIs.

Unified simplifies this by providing:

  • normalized APIs across file storage, messaging, knowledge management, and ticketing platforms
  • consistent objects for documents, messages, and knowledge content
  • unified authentication through Unified Connect
  • webhook events for real-time data updates

Instead of maintaining dozens of integrations, product teams can focus on building AI features that help customers discover and understand their data.

Start building AI search across Google Drive, Slack, Zendesk, Notion, and many other platforms today.

→ Start your 30-day free trial

→ Book a demo

All articles