đŸ› ī¸
Developer docs
Start BuildingGuides
  • ✨Getting Started
  • đŸŽ›ī¸Self Hosted Config
  • âœī¸SSO Login
    • Saml Identity Provider (Idp)
    • Oidc Identity Provider (Idp)
  • đŸŽžī¸Framework Specific Guide
    • âš›ī¸Reactjs
    • âš›ī¸Nextjs
    • âš›ī¸Vuejs
    • âš›ī¸Angular
    • âš›ī¸Svelte
    • âš›ī¸Solid
    • âš›ī¸Vanilla JS
  • â„šī¸Token
  • â„šī¸How to embed?
  • đŸ›ī¸Multi-Tenant Access Control
  • Embed using iFrame (Not Recommended approach)
  • 🔑License Key Validation for Self-Hosted App
  • Test
  • 👩‍đŸ’ģHelpers
    • âœŗī¸Token Body
    • ✅Options
      • Custom Fiscal Year filter setup in DataBrain
    • đŸˆ‚ī¸Server Event
    • Embed Functions
    • Override Language
    • âœˆī¸Embedding Architecture
    • âœˆī¸LLM Architecture
    • ✨LLM Connectors
      • Open AI
      • Claude AI
      • Azure Open AI
      • Llama
      • Mixtral
    • 🆔Workspace Name
    • 🆔Dashboard ID
    • 🆔Metric ID
    • 🆔API Token
    • 🆔End User Metric Creation
    • Embedding APIs
      • Sync Datasource
  • Metric App Filter
  • Dashboard App Filter
  • Chat Mode
    • Step 1: Create Datamart and Workspace
    • Step 2: Create Data App and Embed ID
  • ✨Solutions Alchemy
    • Dashboards for Client Groups
    • Dashboard for Multiple Clients
    • Embedding: Role based Dashboard Filtering
    • Localized Currency Symbols
    • Manage Metrics
Powered by GitBook
On this page
  1. Helpers

LLM Architecture

Architecture of flow of information.

PreviousEmbedding ArchitectureNextLLM Connectors

Last updated 3 months ago

Process Flow

  1. Creation of Datamart on Customer's Server

    • The customer initiates the process by creating a Datamart, specifying tables, relationships, and other data structures on their own server, emphasizing the commitment to data privacy and security.

  2. Confirmation of Datamart Creation

    • A confirmation of the created Datamart is sent back to the customer, signaling readiness for queries.

  3. Query Input by Customer

    • The customer inputs a query with specified dimensions and measures.

  4. Request and Provision of Datamart Schema

    • The customer's server requests the schema of the Datamart from Databrain, which then provides it. This schema is essential for understanding the database structure for subsequent query generation.

  5. Autocompletion and Matching of Query Parameters

    • Databrain employs proprietary AI/ML logic, along with open-source licensed LLMs (Large Language Models), to autocomplete and match the dimensions and measures specified by the customer.

  6. Generation and Verification of SQL Query

    • Databrain generates an SQL query using Data Definition Language (DDL) and labeled information through AI/ML logic.

    • The generated SQL query undergoes a verification process using the same AI/ML logic to ensure its validity.

  7. Execution and Display of Query Results

    • If the SQL query is validated, Databrain executes the query on the customer's server.

    • Query results are then displayed to the customer.

    • In case of an invalid SQL query, an error is reported back to the customer.

Notes

  • AI/ML in Query Generation: Databrain incorporates advanced AI/ML capabilities for intelligent and accurate query generation and validation.

  • Data Privacy and Security: The process is engineered to operate within a secure environment (cloud or VPC), maintaining the integrity and confidentiality of the customer's data.

👩‍đŸ’ģ
âœˆī¸