Overview

This diagram illustrates the architecture of Frontnow’s AI-powered advisory platform, focusing on secure data processing, privacy, and trust. It shows how data flows from various sources through the Frontnow system, integrating cloud-based large language models (LLMs) with robust privacy controls, to deliver AI-generated answers to users.

Main Components & Data Flow

1. Data Ingestion

• Sources:
  • Public Crawling: Collects information from publicly available sources.
  • Live Data Feeds: Pulls in real-time or regularly updated data.
• Importer Module:
  • Both data streams enter the Frontnow environment through an Importer, which ingests, processes, and stores data within the Frontnow EU-Cloud.

2. Frontnow Trust Layer (Core Processing)

• AI Search:
  • Central component for searching and retrieving relevant information from the ingested data.
• Guard:
  • Ensures safety, quality (hallucination scoring), and compliance.
  • Receives feedback from humans for continuous improvement.
• Studio:
  • Likely used for data management, annotation, or curation tasks.

3. Model Integration (Azure Cloud)

• LLM Model(s) (Azure):
  • LLMs hosted on Microsoft Azure with Opt-out and no Training, ensuring customer data privacy.
  • Frontnow IP: Communication with Azure LLMs is done via secure, dedicated Frontnow IPs.

4. Customer Trust Layer+ (Privacy & Compliance for Customers)

• Advisor AI:
  • The core advisory engine delivering AI-powered answers.