Accurate electricity consumption forecasting requires up-to-date and trustworthy weather data. The process is essential for utilities, businesses and cost operators. Yet, the forecasting process comes with significant data infrastructure demands.
a-Gnostics, a company dedicated to crafting A-powered predictive analytics solutions, takes this challenge seriously. Their SaaS solution, Pro-gnostics, was designed specifically to provide forecasting for key energy markets like IESO (Independent Electricity System Operator), MISO (Midcontinent Independent System Operator), and PJM (PJM Interconnection). In order to provide accurate and reliable data, Pro-Gnostics must process large volumes of data quickly, from 40 different locations, reliably, and in as close to real-time as possible.
The location redundancy for weather forecast providers ensures uninterrupted services, even when a provider experiences downtime. But this comes with significant data processing demands. On average, the platform processes about 1 GB of weather data daily — a classic example of data-intensive workloads that demand both speed and efficiency.
Challenges with the previous infrastructure
Before migrating to Tower, the Pro-gnostics Weather Service ran on Microsoft Azure with Docker-based deployments. While reliable, this data engineering infrastructure had significant drawbacks:
High monthly cloud infrastructure costs.
Long execution times, creating inefficiencies.
Forecast runs scheduled only every 3 hours, limiting real-time responsiveness.
The setup worked but fell short of the performance and cost goals a-Gnostics wanted to achieve. They started looking for alternative solutions to run Python apps quickly in a serverless environment - and found Tower.
Migrating to Tower for scalable Python workloads
The migration to Tower was completed in just a couple of days. Existing Python scripts ran seamlessly without modification, enabling a fast migration. The quick move unlocked the full potential of Tower’s scalable Python workloads on a platform built for data engineers.
The impact was immediate:
50% lower infrastructure costs through cloud cost optimization.
70% faster execution times thanks to serverless Python workloads, allowing for near real-time weather inputs.
Retention of a 98%+ accuracy across all forecasts.
By enabling real-time data pipelines, Tower made it possible for a-Gnostics to forecast closer to actual consumption times, boosting accuracy where it matters most and providing a competitive edge on the market.
Business value in energy forecasting
These improvements went beyond technical gains, they created measurable business value. In energy markets like IESO, where financial outcomes are tied directly to forecast accuracy, small improvements have a big impact.
Better forecast accuracy: Fresher data inputs improved model precision.
Financial benefits: Reduced balancing costs and improved trading decisions.
Operational efficiency: Forecasts became both faster and more cost-effective.
The result: a competitive advantage for a-Gnostics on the market and a more reliable system for their customers.
Tower as a cost-effective cloud alternative
Migrating to Tower transformed a-Gnostics’ forecasting infrastructure. What started as a need for faster and cheaper operations became a journey of innovation, efficiency, and business impact driven by data.
Read more about how a-Gnostics Improved Electricity Consumption Forecasts with a Faster and More Cost-Effective Weather Service and Tower.dev