Client: National Claims Funding
Industry: Insurance Receivables Factoring
Website: natclaims.com
Overview
National Claims Funding (NCF) is a specialty finance company that provides non-recourse invoice factoring for contractors and restoration companies. Their clients ranging from roofing contractors, water mitigation specialists, to general restoration firms, perform work on insured properties and then wait weeks or months for the insurance carrier to pay. NCF purchases those receivables upfront, giving contractors immediate cash, and then collects directly from the insurer.
Before funding any invoice, NCF must call the insurance company to verify coverage, claim status, payment history, and adjuster information. That call was being made manually — every single time — by their team.
Prospects Hive designed and deployed a fully automated AI voice agent pipeline that now handles every pre-funding verification call, parses the results, and updates Salesforce automatically. The team no longer dials insurance companies. The AI does it, around the clock, without variation.
The Challenge
NCF’s verification process was both a bottleneck and a risk.
For every invoice, the team had to:
- Call insurance companies manually
- Navigate phone trees and transfers
- Extract claim details during live conversations
- Log notes in Salesforce afterward
Each call took 10-25 minutes.
The issues weren’t just inefficiency, there were:
- Inconsistent underwriting quality
No standardized script meant key details were sometimes missed - Manual data entry errors
Notes varied by agent and were often incomplete - No audit trail
No recordings meant disputes couldn’t be verified - Scaling required hiring
More invoices meant more staff making more calls - Follow-ups were unreliable
Callback requests were frequently lost
The core problem:
Verification quality depended on human execution, not a system.
Our Objective
Build a system that removes the human bottleneck from verification, without reducing accuracy or depth.
Specifically:
- Replace manual calls with a structured AI agent
- Capture all underwriting data points consistently
- Automate CRM updates with zero manual input
- Create a system that scales with volume, not headcount
What We Built for NCF (Step by Step System Overview)
We designed a fully automated verification engine that connects intake, calling, data processing, and CRM updates into one continuous workflow.
1.Intake & Data Capture
The contractor completes NCF’s funding application, uploading their invoice, signed work authorization, policy details, and damage photos. On completion, a Make scenario auto-detects the signed documents and files them to the correct Google Drive folder, organized by claim, and updates the Salesforce record. Zero manual filing.
2.Automated Call Queue
Each morning, a scheduled Make scenario queries Salesforce for all open invoices awaiting pre-funding verification. It pulls the claim data including insured name, address, policy number, claim number, service provider, date of service, and loads each record into Vapi’s outbound call queue for the day.
3.AI Voice Verification
AI agent calls insurance companies automatically Executes a 12-step verification protocol covering: claim validation, coverage status, payment details, and adjuster information.
Handles: Hold queues, transfers, refusals, and callback requests. Every call follows the same structure. No variation.
4.Real-Time Data Processing
The moment the call ends, Vapi sends a structured end-of-call report via webhook to Make.com. The payload includes the full transcript, stereo and mono recording URLs, AI-generated call summary, structured output fields (coverage status, payment status, adjuster details, flags), success evaluation, and all variable values captured during the call.
5.Automatic CRM Updates
The LIVE Vapi Webhook scenario receives and parses the JSON payload in real time. It identifies the correct Salesforce invoice using the invoice ID embedded in the call metadata at launch, then routes the structured data to the CRM update step.
6.Salesforce CRM is updated automatically
Make creates a Salesforce ContentNote on the invoice record containing the full structured call summary, a clickable recording link, short call summary, and all underwriting flags (payment made early, coverage denied, policy lapsed, adjuster refused, etc.). The note is attached directly to the invoice, filed to the right record, instantly, every time.
7.Human calls transcribed and filed via RingCentral + Gemini
For escalations still handled by staff, RingCentral records the call automatically. A Make scenario detects new recordings in the Drive folder, triggers Gemini AI to extract structured data from the transcript, and files the processed output back, maintaining the same data quality standard as the AI calls.
8.Invoice summary emails distributed to the team
A final Make scenario aggregates claim verification statuses from Salesforce on a schedule, groups them by claim, builds formatted summaries, and sends them to the relevant team members, giving NCF a consolidated pipeline view without manual reporting.
Conclusion & Results
Quantitative Wins
- Verification calls reduced from 100% manual → ~0% manual
- Time per verification reduced from 10–25 minutes → 0 minutes of staff time
- Call script consistency improved from agent-dependent → identical every call
- CRM updates improved from manual, post-call → automatic within <10 seconds
- Recording & audit trail improved from none → 100% of calls recorded and linked
- Call capacity scaled from limited by staff hours → unlimited, 24/7
- Scaling shifted from hiring more staff → system configuration-driven
Qualitative Wins
- Underwriting confidence improved: every funded claim is verifiably checked, not relying on staff memory or attention on a given day
- Dispute resolution is now possible: when an insurer disputes what was confirmed on a call, there’s a recording and transcript to reference
- Growth is no longer a hiring problem: NCF can double invoice volume without adding a single verification role
- The ops team works on higher-value tasks: staff shifted from call logging to actual underwriting analysis and client management
Tech Stack Used
- Vapi — AI voice agent for outbound calls.
- Make.com — automation orchestration layer.
- Salesforce — CRM and system of record.
- RingCentral — human call recording.
- Google Gemini AI — transcript processing and data extraction.
- Google Drive + PandaDoc — document management.
All systems are connected through Make.com, creating a unified workflow.
What They Had To Say
Want to automate a process like this?
We design and build AI automation systems for operations-heavy businesses. If your team is doing repetitive, high-stakes work that could be handled by an intelligent agent, let’s talk.
Book a strategy call and see what your operation looks like as a system.
