Gecko Alliance

A 30-year technology partner to the spa industry

Gecko Alliance Group is a Canadian company that designs and manufactures technology for the spa and hot tub industry. For more than 30 years, it has developed control systems, keypads, pumps, touchscreens, remotes, mobile applications and accessories for spa manufacturers, distributors, retailers and technicians worldwide.

Headquartered in Quebec City, Gecko has more than 450 employees across four locations and holds ISO 9001 and CTPAT certifications.

Reading a test strip by eye is slow and subjective

Reading a spa test strip by eye can be slow, subjective and error-prone. Lighting conditions, glare and pad alignment can all affect how colors are perceived and make manual interpretation inconsistent.

As a technology partner to spa manufacturers, Gecko wanted to let spa owners obtain an accurate water chemistry reading from a single phone photo, directly within its connected spa ecosystem and without requiring dedicated reading hardware.

The challenge was not simply recognizing colors. The solution also had to handle the variability of real-world photos while avoiding confident but incorrect readings.

Multimodal vision for perception, deterministic logic for chemistry

Ingeno designed and built a serverless analysis service on AWS.

A spa owner takes a photo of the test strip in the mobile application. The image is uploaded securely to Amazon S3, then a foundation model on Amazon Bedrock uses native multimodal vision to identify structured color information for each pad. A deterministic processing step maps those observed colors to the reference chart and produces the final water chemistry reading.

This separation keeps the foundation model focused on visual perception while the conversion from color to chemistry values remains deterministic, consistent and auditable.

Ingeno also implemented quality safeguards for real-world conditions, including a confidence score for each pad, reason codes, a pad-count validation and a color-distance threshold. When a reading is uncertain, the service can flag it rather than return an unreliable result.

Because the service is built on a serverless architecture, it can scale automatically with demand while keeping the operational footprint lightweight.

From about 30 percent to about 90 percent automated success

The service is currently in production inside Gecko’s connected spa ecosystem.

Automated analysis success rate increased from about 30 percent to about 90 percent.

Mobile app adoption grew from about 500 to about 8,000 users following deployment of the new experience.

Spa owners can obtain a water chemistry reading from a single phone photo without dedicated reading hardware.

Confidence scoring and validation safeguards help prevent uncertain readings from being presented as reliable results.

The serverless architecture scales on demand with minimal operational overhead.

Using a foundation model also allowed the solution to reach production faster than a traditional computer vision approach. It did not require a labeled training dataset and could handle variations in lighting through prompt design and per-pad confidence scoring.

Amazon Bedrock, AWS Lambda, Amazon S3, AWS Systems Manager Parameter Store and Amazon CloudWatch.