Technology

Technical architecture and AI models

System Architecture

High-level overview of Faxi's components and data flow

Marketing Website (Astro) Home / Service / Demo / Tech / Help HTTP/REST Backend API (Express.js) Demo Endpoints / Metrics API / Webhook Handlers AI Processing Pipeline Vision AI OCR / Handwriting Annotation Detector Checkmarks / Circles Intent Extractor Action Classification MCP Servers Email / Shopping / AI Chat / Payment / Appointments Infrastructure PostgreSQL Data Storage Redis Queue & Cache S3 File Storage Telnyx Fax API

AI Models & Techniques

State-of-the-art AI powering accurate fax interpretation

Multi-Model AI Pipeline

Faxi uses a sophisticated AI pipeline combining multiple specialized models. Each model excels at a specific task, and their outputs are combined to achieve high overall accuracy.

Vision AI (GPT-4 Vision)

Optical Character Recognition and Visual Analysis

92%
Accuracy

Extracts text from fax images including both printed and handwritten content. Uses advanced computer vision to understand document structure, identify form fields, and recognize Japanese characters with high accuracy.

Multimodal deep learningTransformer architectureHandwriting recognitionForm field detectionLayout analysis

Annotation Detector

Visual Annotation Recognition

88%
Accuracy

Identifies hand-drawn marks on faxes such as checkmarks, circles, arrows, and underlines. Associates annotations with nearby text to understand user intent.

Convolutional neural networksEdge detection algorithmsShape recognitionConfidence scoringBounding box regression

Intent Classifier (Claude)

Natural Language Understanding and Action Extraction

95%
Accuracy

Analyzes extracted text and annotations to determine what action the user wants to perform. Classifies intents (email, shopping, appointment, etc.) and extracts relevant parameters with high confidence.

Large language modelsFew-shot learningContext-aware parsingEntity extractionSemantic analysis

Processing Pipeline

1

Image Preprocessing

Enhance image quality, remove noise, correct skew and rotation

Gaussian blurAdaptive thresholdingMorphological operations
2

Vision Analysis

Extract text regions and identify visual elements

OCRLayout detectionHandwriting recognition
3

Annotation Detection

Find and classify hand-drawn marks

Shape detectionPattern matchingSpatial analysis
4

Intent Extraction

Understand user intent and extract parameters

NLPEntity recognitionContext analysis
5

Confidence Scoring

Assess reliability of each component

Ensemble methodsUncertainty quantificationValidation checks

Overall Performance

90%+
Overall Accuracy
<5s
Avg Processing Time
95%+
Intent Classification
10+
Supported Use Cases

Technology Stack

Modern, scalable technologies powering the Faxi platform

Frontend

Astro

Fast static site generator for optimal performance

TypeScript

Type-safe JavaScript for robust code and better developer experience

Tailwind CSS

Utility-first CSS framework for rapid UI development

Backend

Express.js

Fast, minimalist web framework for Node.js

PostgreSQL

Robust relational database for storing users, jobs, and metrics

Redis

In-memory data store for job queues and caching

AI & Machine Learning

Claude (Anthropic)

Advanced language model for intent extraction and NLU

GPT-4 Vision

Multimodal AI for OCR, handwriting recognition, and visual analysis

Infrastructure

Telnyx

Cloud communications platform for sending and receiving faxes

Docker

Containerization for consistent development and production environments

MCP Integration

Model Context Protocol servers extend Faxi's capabilities

What is MCP?

Model Context Protocol (MCP) is an open standard that enables AI systems to securely connect with external data sources and tools. Faxi uses MCP servers to extend functionality beyond basic fax processing.