Cobalt Intelligence Lender Workflows & Use Cases – Real-Time Application Verification

The Problem: The "Fake Business" Funnel

Every high-volume lender faces the same operational drain: underwriting teams spending 20-30 minutes manually verifying businesses that don't exist. Fraudsters submit applications with made-up names like "Prestige Capital LLC" in Delaware, knowing that by the time a human underwriter manually checks the Secretary of State (SOS) website, the application has already moved through initial staging. This wastes costly underwriting hours on applications that should have been auto-rejected in milliseconds.

The Solution: Instant SOS Verification at Intake

Shift the verification step left. Instead of waiting for an underwriter to pick up the file, trigger a live Cobalt Intelligence SOS API call the second the applicant clicks "Submit." This allows you to auto-reject dissolved, inactive, or non-existent entities before they ever enter your loan origination system (LOS).

Expected Outcome:

• Auto-Reject: Fraudulent or "Dissolved" entities in <5 seconds.
• Cost Savings: Reduce manual SOS lookups by ~40-60%.
• Speed: Active businesses move to credit pull immediately.

Workflow: Real-Time Verification

Integration Point:

The API trigger typically sits between your Application Form (Intake) and your LOS / CRM.Getty ImagesExplore

1. Applicant Input: Submits Legal Name ("Acme Corp"), State ("TX"), and EIN.
2. API Trigger: System calls Cobalt SOS API using Name + State.
3. Status Check:
   
   • Is the status "Active" or "Good Standing"? → Pass to Credit Pull.
   • Is the status "Inactive", "Forfeited", or "Dissolved"? → Auto-Reject.
4. Confidence Check:
   
   • Is the match confidence < 0.80? → Flag for Manual Review (Potential typo or DBA).

Technical Implementation

1. API Call: Entity Search

You typically verify the entity using the businessName and state. While you collect the EIN for tax verification (TIN Check), the SOS lookup relies on the entity name and jurisdiction.

Code Snippet (Node.js Example):

JavaScript

const axios = require('axios');

‍

async function verifyBusiness(businessName, state) {

  const apiKey = 'YOUR_COBALT_API_KEY';

  const url = `https://apigateway.cobaltintelligence.com/v1/search`;

‍

  try {

    const response = await axios.get(url, {

      headers: { 'x-api-key': apiKey },

      params: {

        searchQuery: businessName, // e.g., "Acme Innovations LLC"

        state: state,              // e.g., "TX"

        liveData: true             // Force real-time SOS lookup

      }

    });

‍

    const result = response.data.results[0]; // Get top match

    return evaluateEntity(result);

‍

  } catch (error) {

    console.error("Verification failed:", error);

    return "MANUAL_REVIEW";

  }

}

2. Decision Logic: The "Knockout" Rules

Don't just store the data—act on it. Use the status and confidenceLevel fields to drive automated decisions.

JavaScript

function evaluateEntity(entity) {

  // 1. No Match Found

  if (!entity) {

    return "AUTO_REJECT: No Record Found";

  }

‍

  // 2. Status Check (Auto-Reject bad standings)

  const badStatuses = ['Dissolved', 'Inactive', 'Forfeited', 'Suspended'];

  if (badStatuses.includes(entity.status)) {

    return `AUTO_REJECT: Status is ${entity.status}`;

  }

‍

  // 3. Confidence Check (Flag typos/fuzzy matches)

  if (entity.confidenceLevel < 0.80) {

    return "MANUAL_REVIEW: Low Confidence Match";

  }

‍

  // 4. Success

  return "APPROVE: Active Entity Verified";

}

Decision Tree: Match Confidence Thresholds

Cobalt API returns a confidenceLevel (0.0 to 1.0) indicating how closely the SOS record matches your applicant's input. Set your thresholds to balance speed vs. risk.

• Score 0.90 - 1.00 (Exact/High Match):
  
  • Action: ✅ Auto-Approve.
  • Context: Name matches perfectly or with minor suffix variations (e.g., "Inc" vs "Inc."). Proceed to credit underwriting.
• Score 0.80 - 0.89 (Good Match):
  
  • Action: ✅ Auto-Approve (with address validation).
  • Context: Strong match but maybe a slight spelling difference. If the Zip Code also matches, approve it.
• Score < 0.80 (Low Confidence):
  
  • Action: ⚠️ Manual Review.
  • Context: Significant name variation. Could be a DBA ("Joe's Pizza" vs "Joseph Smith LLC") or a different entity entirely. Do not auto-reject, but do not auto-approve.

Automating Loan Workflow with Cobalt Intelligence

This video demonstrates how accessing real-time Secretary of State data, specifically officer information, fits into a fraud prevention workflow to verify who really owns the business.

‍