# Custom algorithm | ePhi Case Study Canonical: https://www.flobotics.io/case-studies/custom-algorithm-for-ephi Published: 2026-04-08 Updated: 2026-04-08 --- [Home](https://www.flobotics.io/)/[Client Results](https://www.flobotics.io/case-studies)/Custom algorithm | ePhi Case Study Business process automation (Non-healthcare) # Custom algorithm | ePhi Case Study See how we delivered an algorithm capable of capturing, organizing, and sharing satellite images with consistent scale, definition, and predictable costs for ePhi - industrial cleaning services company from France. ClientePhi WorkflowBusiness process automation (Non-healthcare) On this page - [Client](#client) - [The Challenge](#the-challenge) - [The Solution](#the-solution) - [The Outcome](#the-outcome) Book a callWant results like these? 30 minutes with someone who builds automations, not a sales script. [Talk to us →](https://www.flobotics.io/contact) Share [](#)[](#) ## Client ePhi is an environmental services company based in France, providing industrial cleaning services. ## The Challenge ePhi needed to process satellite imagery covering large industrialized areas across multiple cities. The goal was to analyze every square meter of land and identify buildings that could be classified as commercial development targets. Due to the massive data volumes involved, manual processing was not feasible. Initial plans to build custom AI solutions in-house proved too costly and time-consuming. ## The Solution After gathering project requirements, Flobotics selected tools that balanced cost, quality, and development speed. The map scraping process was based on Google Maps, which provided an optimal ratio of satellite image quality to cost. Scraping entire metropolitan areas typically cost less than $90,depending on size, and the API ensured consistent image resolution and dimensions. Low development complexity further reduced overall project costs. Because Google Maps Static API is not designed for machine learning use, we developed a custom algorithm to refine and enrich the image data in real time. The automation was deployed on Amazon Web Services, due to API rate limits requiring long-running processes.AWS enabled fire-and-forget execution, and images were stored in AWS S3 for seamless integration with machine learning services such as SageMaker. ## The Outcome Flobotics delivered an algorithm capable of capturing, organizing, and sharing satellite images with consistent scale, definition, and predictable costs. The images were automatically transformed into structured, readable data for future analysis. The entire development process took less than two months, meeting the client’s time constraints. ePhi, thank you for your trust. Let’s talk! It’s a 30-minute call. Within the first 15, we’ll know if automation works for you. [Book a call →](https://www.flobotics.io/contact) More work ## More case studies Real deployments, with the numbers our clients measured. [ Denials Management Up to 80%higher throughput in RCM, with a custom laboratory information system data exchange. ### Up to 80% Higher Throughput in RCM Read the case study → ](https://www.flobotics.io/case-studies/up-to-80-higher-throughput-in-rcm---denials-management-automation)[ Eligibility Checks 100%ROI in 23 days from eligibility automation plus Denial Hunter, with $180,000 in savings after year one. ### 100% ROI in just 23 days. We Automated Eligibility Checks! Read the case study → ](https://www.flobotics.io/case-studies/100-roi-in-just-23-days-we-automated-eligibility-checks) [All case studies →](https://www.flobotics.io/case-studies)