From 25 Minutes of Manual Document Checks to 5-Minute with an AI-Verified Claims Pipeline

The Challenge

Before automation, the intake team had to manage a multi-document process almost entirely by hand. Staff would receive ID images, bills, and bank statements via email or uploads and then visually scan each page to find names, dates of birth, addresses, and transaction amounts. This typically took about 25 minutes of human time per claimant.

As a result, a historic list of around 20,000 leads largely sat untouched; recontacting them and manually checking each case was simply too much work. New leads faced similar problems, often going cold before their cases were properly documented.

The Solution

The engagement focused on building a high-volume intake engine centered on:

• Go High Level (GHL): The CRM and system of record for contacts, fields, and case status.
• n8n: The orchestration layer connecting messaging, OCR, and AI extraction.
• WhatsApp Business: The client-facing channel for simple, guided interaction.
• Mistral-based OCR + AI Extraction: Automatically reading documents and pulling structured data for legal review.

How It Works

From the claimant's perspective, the process feels like a guided, conversational checklist on WhatsApp. They are prompted to upload ID, utility bills, and bank statements one at a time. Behind the scenes, the system handles:

• Message Routing: Listening for messages and extracting media.
• AI Extraction: Converting raw OCR text into structured data (DOB, address, postcodes, etc.).
• Validation: Comparing claimant-stated amounts against bank statement transactions.
• Automatic Flagging: Instantly notifying claimants if a document is unreadable or doesn't match expectations.

Key Outcomes

• 80% less manual checking per clean case.
• Ability to work through 20k historic leads without adding staff.
• Scaleable intake for new Facebook campaigns without overwhelming the team.