Form Mapper/Filler Architecture

AI-Powered Zero-Knowledge Form Completion

Upload once → Map intelligently → Fill securely → Audit transparently

Overview User Flow Technical Metrics Security

Key Features

AI Field Detection

LLM analyzes PDF forms to understand field meanings, types, and requirements

Smart Mapping

Semantic matching suggests vault data for each field with confidence scores

Completion Metrics

Real-time tracking of filled fields, missing data, and overall completeness

End-to-End Process

1

Agent Uploads Form Template

Agent (mec.cx)

Landlord uploads rental application PDF

PDF.js parsing + form field detection
2

AI Form Analysis

Client-Side AI

Extract fields, detect types, understand semantic meaning

LLM + vision model (GPT-4V or Claude) via API
3

Vault Data Mapping

Client-Side

AI suggests mappings: 'Full Name' → user.legal_name

Vector embeddings + semantic matching
4

User Review & Approval

User (mecentral.io)

Review mappings, approve data sharing, see completion %

React UI with field-by-field control
5

Client-Side Form Filling

Client-Side

Decrypt vault data, fill PDF fields, validate completeness

pdf-lib + zero-knowledge decryption
6

Audit & Delivery

Ledger + Agent

Log data elements used (not values), deliver to agent

Immutable audit log + agent notification
~30 sec
Form Analysis Time
95%
Auto-Mapping Accuracy
Zero
Server Data Exposure
100%
Audit Trail Coverage

Transform Your Form Workflow

Never manually fill the same form twice