> ## Documentation Index
> Fetch the complete documentation index at: https://swiftgum.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.

# Architecture

> Overview of knowledge sourcing and system architecture using Mermaid diagrams.

# Knowledge Sourcing & Engine Architecture

## Knowledge Sourcing

<Info>
  Below is the Mermaid diagram representing the Knowledge Sourcing pipeline, explaining how data is ingested, processed, and stored.
</Info>

```mermaid theme={null}
flowchart LR
    %% Define external sources
    AdminKnowledge["Admin Knowledge<br/>(Company websites, Notion, etc.)"]
    UserKnowledge["User Knowledge<br/>(e.g. mails, personal Drive, draft Notion, ...)"]

    %% Subgraph for knowledge sourcing logic
    subgraph Knowledge Sourcing
        direction LR
        Drive["Drive"]
        Notion["Notion"]
        Refresh["Refresh"]
        AccessControl["Access Control"]
        Embeddings["Embeddings"]

        Drive --> Refresh
        Notion --> Refresh
        Drive --> AccessControl
        Notion --> AccessControl
        Refresh --> AccessControl
        AccessControl --> Embeddings
    end

    %% Connect external sources into the subgraph
    AdminKnowledge --> Drive
    UserKnowledge --> Notion

    %% Downstream steps after Embeddings
    Supabase["Supabase<br/>Table w/ Vectors"]
    RAG["RAG"]
    Inference["Inference"]
    Mem0["Mem0"]

    %% Embeddings flow to vector DB + RAG
    Embeddings --> Supabase
    Embeddings --> RAG

    %% RAG leads to Inference
    RAG --> Inference

    %% Inference can use or feed Mem0
    Inference --> Mem0
    Mem0 --> Inference
```

### Explanation

* **Knowledge Sourcing subgraph**: Demonstrates the ingestion pipeline from Drive and Notion through Refresh, Access Control, and Embeddings.
* **Admin/User Knowledge**: External sources feeding into the system.
* **Supabase, RAG, Inference, Mem0**: Shows AI feature pipelines (RAG-based retrieval, inference, and memory functionality).

***

## Engine & App Architecture

<Info>
  The following Mermaid diagram illustrates how the Engine processes files and interacts with the App, which manages integrations, OAuth flows, and storage.
</Info>

```mermaid theme={null}
flowchart LR
    subgraph Engine
        direction TB
        ProcessFiles["Process Files<br/>(e.g., Python to convert to Markdown)"] --> UpdatesDB["Writes new/updated data"]
    end

    subgraph App
        direction TB
        BackendDB["Backend Database<br/>(SQLite local or Supabase cloud)"]
        ManageIntegrations["Manage Integrations + Set up External Sync"]
        OAuthFlows["Handle OAuth Content Flows"]
        ThirdParty["3rd-Party (Google Drive, Notion, etc.)"]
        
        BackendDB -->|"Stores Token"| OAuthFlows
        OAuthFlows -->|"Initiates OAuth"| ThirdParty
        ManageIntegrations --> BackendDB
    end

    Developer --> ManageIntegrations
    Engine -->|"Reads Token/DB String"| BackendDB
    Engine --> ProcessFiles
    UpdatesDB -->|"Push Data"| Supabase["Supabase (or Postgres)"]
    Developer --> UIKit["UI Kit"]
    UIKit --> EndUser["End User"]

    %% Show relation from App to Engine
    App -->|"Provides DB info / Token"| Engine
```

### Explanation

* **Engine**: Processes files (e.g., converts to Markdown) and updates the database.
* **App**: Manages integrations, tokens, and OAuth flows with external providers.
* **Developer**: Configures integrations.
* **UI Kit & End User**: Users interact with UI components to connect their accounts.