Artificial Intelligence for property valuation: A paradigm shift

Cómo desarrollamos una solución digital para la gestión de créditos hipotecarios en Bancor

A global leader in satellite solutions and geospatial analysis set out to transform a key process for municipalities in Colombia: the typological classification of housing units within their property valuation projects.

Traditionally, this process was carried out by human field teams who collected RGB images and manually assigned a construction typology based on a technical dictionary. While functional, this approach had its limitations: it was slow, expensive, and heavily dependent on the individual judgment of each operator.

The organization sought a solution that could ensure consistency, technical precision (with an acceptance standard above 97.5%), and the ability to scale without increasing costs. To achieve this, they turned to Exomindset.

The solution: Artificial Intelligence applied to urban analysis

At Exomindset, we designed a custom solution: a system that automates the typological classification of housing based on images, using an advanced Vision Language Model (VLM).

The tool enables the following:

  • Detects key elements such as roof type, materials, and distinctive structures.
  • Cross-references data with a semantic matrix of typologies, automatically assigning the most likely classification.
  • Delivers customizable reports, with full traceability and ready for integration with the client’s logistics systems.

This entire pipeline runs in a secure and scalable AWS environment, ensuring access control, system compatibility, and maximum operational availability.

Expected results:

  • Automate construction classification with a uniform technical standard.
    Drastically reduce manual analysis and review time.
  • Optimize the use of technical resources by focusing on the most complex cases.
  • Raise the standard of property valuation analysis across Latin America.

A multidisciplinary team tackling a real challenge

This project brought together senior technical profiles across various domains:

  • Delivery Manager
  • Senior ML/AI Engineer
  • Senior DevOps Engineer
  • Senior Web Developer

We worked in close collaboration with the client’s technical team to understand their operational workflows and ensure the solution integrated seamlessly.

The technology behind the transformation

  • Languages & Frameworks: Python, Next.js
  • Cloud Infrastructure: AWS Lambda, EC2 Linux, S3, DynamoDB, Amplify
  • DevOps: AWS CodeCommit, CodeBuild, CodeDeploy, CodePipeline