Skip to content
FelgorFELGOR
Registry cleanup and enrichment•Operations with large customer bases or address-critical points

FELGOR Vision

Cleans address-critical registries by cross-checking multiple sources — no blind trust in AI

The problem

Address-critical registries are the foundation for many things — routing, commissions, billing, customer service, segmentation. And they're always broken: wrong addresses, off-target coordinates, inconsistent typology, duplicates. Each error becomes a wrong route, a missed bill, a service call sent to the wrong place, a miscalculated commission. Auditing thousands of records manually is unfeasible. Auditing with raw AI creates a different problem: the AI decides it knows and makes things up — hallucinates addresses, geocodes in the wrong place, and nobody notices.

The solution

FELGOR Vision eliminates hallucination by construction. Instead of trusting a single AI, it asks the same question to independent sources — Google Maps, OpenStreetMap, the federal tax registry, visual storefront analysis via Street View. When all agree, confidence is high. When they diverge, it flags it for the operator to decide with the evidence in front of them. Everything logged, auditable.

Capabilities

What the product does

  • Receives a registry spreadsheet and processes every record without manual typing
  • Cross-checks address with Google Maps, OpenStreetMap and the tax registry
  • Validates typology (POS, residence, commerce, industry) by combining the registry with visual analysis
  • Detects bad geocoding with high confidence — coordinate that doesn't match the address against any source
  • Detects AI hallucination by cross-checking with human-curated sources
  • Field-specific confidence score — operator knows where to look
  • Auditable report: what was corrected, based on which source, with what confidence

Where it applies

It's not a vertical product

The engine is the same. What changes is the catalog trained for each context. Below are the sectors where we've already seen it make sense — but the use cases don't stop here.

  • Last-mile distribution (POS)
  • Collections and credit recovery
  • Insurance carriers
  • Banks and fintech (KYC, expansion)
  • Healthcare (patient and beneficiary records)
  • E-commerce and reverse logistics
  • Utilities (power, water, gas)

How it works

Day to day

  1. 1

    Operation submits the registry spreadsheet

  2. 2

    System cross-checks every record against multiple sources in parallel

  3. 3

    When sources agree, the update is automatic

  4. 4

    When they diverge, the record goes to a human-review queue with consolidated evidence

  5. 5

    Operator validates, system updates, enriched spreadsheet comes out the other side

Gains

What changes in the operation

  • Operations trust the registry enough to automate what used to be manual
  • Errors caused by rotten data stop appearing at the end of the chain
  • Registry audit becomes a process, not a once-a-year event
  • AI hallucination risk is eliminated by architecture, not by policy

Who this is for

Any operation with a large customer base or set of points where address feeds critical processes. Sector doesn't matter — the engine cross-checks address, coordinate and typology regardless of what the registry describes. A beverage distributor uses it for points of sale, a bank for branches and clients, an insurer for insured residences, a hospital for patients.

Works best when:

  • The base has 10k+ records (below that, manual audit is feasible)
  • The registry feeds a critical process (route, billing, service, segmentation)
  • There's a known gap between the registry and reality

If the address is just for regulatory compliance and nobody operates against it, the product has no return.

What FELGOR Vision isn't

Not a pure geocoding service. It geocodes, but as one of the sources — and never trusts a single one. Doesn't promise a 100% perfect registry. Promises to make what's wrong visible, with calibrated confidence. Doesn't replace the new-record creation process. Cleans the legacy and maintains the existing one — new records keep entering through the existing channels.

How it fits in

Runs as a service over the existing registry spreadsheet. Output is an enriched spreadsheet plus an auditable report. No CRM or ERP swap. Can be one-off (initial cleanup) or recurring (continuous audit).

Want to learn more about FELGOR Vision before booking?

Read articles on this solution

Got a problem similar to FELGOR Vision in your operation?

In 30 minutes with the founders, we'll map your context and tell you whether an adaptation makes sense — or whether a different approach fits better. Straight talk.

Book a diagnosis for my case

30 min · free · with the founders