Automate fraud detection and data cleaning for LATAM with n8n

Built with n8n
, ,
1

We built a community node for n8n that connects to our API — focused on Latin American data problems that generic tools don’t handle well.

Install

npm install n8n-nodes-mediavox

Or search “Mediavox” in the n8n community nodes UI.

What it does (48 operations)

Fraud detection in workflows

Trigger: incoming webhook with SMS/WhatsApp message → Mediavox analyzes it → route based on verdict.

Example: a fintech receives forwarded suspicious messages from users. The workflow:

  1. Webhook receives the message
  2. Mediavox Threat Analyze checks brand impersonation, URL safety, urgency patterns
  3. IF verdict = “fraudulent” → alert user + log to DB
  4. ELSE → respond “looks safe”

Works with 263+ LATAM brands (banks, telcos, fintechs across CO, MX, PE, CL, EC, AR).

Data cleaning before CRM import

Your CSV has names like “INVERSIONES DEMO S.A.S.” and phones like “3001234567” (no country code). Before loading into your CRM:

  1. Read CSV node
  2. Mediavox Names Standardize → splits company name + legal suffix + detects country
  3. Mediavox Emails Validate → catches disposables, typos, dead MX records
  4. Mediavox Phones Validate → normalizes to E.164, flags country mismatches
  5. Write to CRM with clean data

Sanctions screening on new suppliers

Every new supplier added to your system gets screened automatically:

  1. Database Trigger on new supplier row
  2. Mediavox Sanctions Check → screens against OFAC, UN, EU, PEP lists with Spanish fuzzy matching
  3. IF match found → flag for compliance review
  4. ELSE → mark as cleared

Handles accents, name order variations, and abbreviations that English-only tools miss (“José García” vs “GARCIA LOPEZ, JOSE”).

Available operations

Category Operations
Security Threat analyze, sanctions check, brand monitor, threat feed
DataTools Names standardize, emails validate, phones validate, addresses parse, domains check, quality score
Recognition OCR extract, document analyze
Finance Tax ID validate, bank account verify

Setup

  1. Get a free API key (no credit card) from the developer portal
  2. Add credentials in n8n: Settings → Credentials → Mediavox API
  3. Drop the node into any workflow

Free tier: 500 requests/month across all operations.

Built specifically for Spanish-speaking LATAM. If you’re automating workflows for that region and hitting walls with English-focused tools, give it a shot.

3 Likes
2

Hi @mediavox,

Kudos on launching this! As someone who builds a lot of enterprise AI pipelines and data integrations, I can confidently say that data normalization for the LATAM region is a massive pain point. English-centric tools completely fall apart here.

A few aspects of your node particularly stand out from an architecture perspective:

  • Spanish Fuzzy Matching: Standard LLMs or basic regex algorithms fail miserably when matching names like "José García" vs "GARCIA LOPEZ, JOSE" against OFAC/PEP compliance lists. Having native handling for double surnames, accents, and local abbreviations directly inside an n8n node is a huge time-saver.

  • Phone & Tax ID Normalization: Cleaning up raw data (like adding local country codes to match E.164 formatting) before importing it into CRMs or DBs prevents massive data debt down the line.

To help me understand how to best position this in production-grade enterprise workflows, I have two quick technical questions:

  1. Tax ID & Bank Verification: Does the Tax ID validate operation perform real-time checking against live government databases (e.g., SAT in Mexico, DIAN in Colombia, AFIP in Argentina), or is it currently limited to offline syntax/checksum algorithms?

  2. Rate Limits & Scaling: For high-throughput webhooks (like real-time fraud detection on incoming traffic), how does the API handle sudden spikes? If we scale past the 500-request free tier, what do the latency benchmarks look like for the Threat analyze operation?

Excellent contribution to the n8n ecosystem. I’ll definitely be giving n8n-nodes-mediavox a spin on my next LATAM-focused automation pipeline!

3

Thanks @Swapnil_Mandloi — appreciate the detailed questions. Let me address both:

1. Tax ID Validation — live vs offline:

Currently it’s syntax + checksum validation (offline algorithms). This covers format verification, check digit calculation, and structure rules per country:

  • CO (NIT): modulo-11 check digit
  • MX (RFC): homoclave + valid date encoding
  • PE (RUC): prefix rules + check digit
  • CL (RUT): modulo-11
  • EC (RUC/CI): provincial prefix + check digit

We don’t hit live government endpoints (SAT, DIAN, AFIP) because: (a) most don’t have public APIs, (b) those that do have aggressive rate limits and require taxpayer authorization, (c) for onboarding workflows you typically need instant response (<500ms), which live lookup can’t guarantee.

That said — if the ID passes format + checksum, it’s structurally valid. False positives (valid format but non-existent entity) are rare in practice because the algorithms reject most random inputs.

2. Rate limits & scaling:

  • Free tier: 500 req/month (for testing)
  • Paid plans: 5K–100K+ req/month depending on tier
  • Threat Analyze latency: p95 < 800ms (includes URL resolution if shortened links are present; without URL hops it’s ~200ms)
  • Burst handling: we allow short bursts up to 50 req/s before throttling kicks in (429 with Retry-After header)
  • No queue — all requests are synchronous. If you need to handle spikes beyond your plan’s rate, the node gracefully returns the 429 and n8n’s built-in retry mechanism handles backoff.

For high-throughput production (real-time webhook fraud detection), the Scale plan at 100K req/month handles ~2.3 req/s sustained, which covers most fintech inbound volumes. If you need more, reach out — we can discuss dedicated capacity.

Let me know how the testing goes. Happy to help with any integration edge cases.