How We Turned Fax Hell Into a Fully Searchable Database in 2 Days

Every healthcare org has a dirty secret: the fax machine is still running the show.

Whether it’s incoming referrals, prior auth requests, medical records, or insurance forms — they all come in via fax. Often scanned, rotated, low-res, handwritten, and totally disconnected from your EHR or billing platform.

You can't search it. You can't track it. You can barely read it.

We were tired of that. So we built a system to fix it in 48 hours.

Here’s how we turned a chaotic inbox of unstructured faxes into a structured, searchable, API-connected database — using Nanonets AI agents.

The Problem: Death by Fax

Clinic setup: 5-provider ENT practice, high-volume referrals Incoming faxes/day: 60–90 Team process:

• Print or download PDF from e-fax inbox
• Read manually
• Copy data into EHR
• Rename file, move to folder

“We had no way to search old faxes. If someone asked, 'Did we get Dr. Patel's referral for Jane Doe?', we had to scroll through 300 files manually.” — Practice Admin

The Solution: Build a Searchable Fax Engine

We deployed a two-part system:

1. Nanonets OCR + NLP agent

• Watched fax inbox (PDFs from email)
• Parsed each document in real-time
• Extracted key metadata: patient name, DOB, referring provider, referral reason, document type

2. FastAPI + Postgres DB

• Stored each fax with structured metadata
• Searchable by name, provider, keywords
• Accessible via internal dashboard or API

All deployed in <2 days.

What It Looks Like

Before:

• Folder of cryptic PDF names: referral_12345.pdf, scan_02-17.pdf
• No visibility or audit trail

After:

• Each fax has a searchable entry:

{
  "patient_name": "Jane Doe",
  "dob": "1990-05-19",
  "referring_provider": "Dr. Patel",
  "document_type": "Referral",
  "date_received": "2024-02-16",
  "original_file": "referral_jane_doe.pdf"
}

• Staff can:
  • Search: "referrals from Dr. Lee last week"
  • Filter: "all prior auth forms for Blue Cross"
  • Export: "all faxes received this month"

Why It Worked

1. AI that understands faxes

• Trained on thousands of provider layouts
• Handled low-light scans, handwritten forms, rotated pages

2. No change to staff workflow

• They still used the same fax inbox
• System ran in background, no learning curve

3. Lightweight deployment

• No EHR integration required
• FastAPI backend + simple UI

4. Immediate time savings

• Manual data entry dropped by 70%
• Staff saved 2–3 hours/day

Impact Metrics

“The ability to search any fax by patient name or provider changed everything for our team.” — Front Desk Lead

Bonus: Analytics + Alerts

Now that faxes were structured, we added:

• Referral source tracking: Which doctors send the most referrals
• Daily volume alerts: Know if fax flow dips unexpectedly
• Flagged documents: Auto-detect missing info or bad scans

You Can Do This Too

Here’s what you need:

• E-fax inbox (email-based preferred)
• Nanonets OCR agent access
• Simple DB (Postgres) + web framework (FastAPI, Flask)
• 1 engineer + 1 day

Optional: dashboard, webhook triggers, document tagging

TL;DR

Faxed documents are still the norm. But leaving them unstructured is costing you:

• Time
• Revenue
• Visibility

With the right AI agent and lightweight infra, you can:

• Parse every fax
• Search everything
• Route docs instantly

No more scroll-and-hope.

Want to see a live demo of the searchable fax inbox? Talk to our team. We’ll show you exactly how we built it, and for a small fee we can build one out for you too.

Sources:

• Internal Nanonets deployment, ENT clinic (2024)
• KLAS Research: EHR Interoperability & Fax Usage Trends (2023)
• MGMA: Staff time spent processing inbound documents