Smart Light uses IBM watsonx.ai to create a new gen AI workflow for facilities control
Facilities control has IoT aspects that differ depending on the specific situation of each space. Conventional dialogue flows already present in smart speakers have difficulty adapting to various situations. In addition, it has been challenging to use new technologies such as generative AI (gen AI), which have recently become popular, in this scenario.
Since facilities control has a variety of protocols, Smart Light, a Japanese company specializing in integrating electric systems, needed a technology bridge that could effectively handle voice and gen AI coordination.
To start, Smart Light worked on the development of efficient dialogue flows. The company chose a method in which the gen AI tool would understand the context of the text transcribed from the voice. It also relied on 1ft-seabass, an independent technology studio, for technical support, overall architecture planning and integration of IBM products.
Since there weren’t many examples available in the market, Smart Light had to experiment and verify the newly created gen AI workflows. Smart Light used the IBM® watsonx.ai® enterprise-grade AI studio as the core of its solution and focused on a system that allowed trial and error using a graphic user interface (GUI).
The IBM Cloud® Code Engine serverless platform was used for equipment control protocol conversion. Its interface was standardized to the Message Queuing Telemetry Transport (MQTT) messaging protocol by using the Node-RED development tool. Moreover, the on/off context was interpreted by watsonx.ai and bridged by IBM Cloud Code Engine.
The advantages of integrating AI and facilities control soon became clear. Smart Light, with the support of 1ft-seabass, was able to implement the gen AI voice interpretation solution effectively—an excellent step toward the future of AI and facilities control cases. Specifically, in the test room, the company was able to manage the air conditioning device using voice, and this prototype was quickly implemented in about two weeks.
In addition to air conditioning control, Smart Light also achieved equipment control—such as lighting and curtain control—by connecting Node-RED and Code IBM Cloud Engine using MQTT. There were several trials when testing the AI prompts in the playground, and being able to try them out repeatedly helped speed up the development of the new solution.
In the past, it would have taken a week to complete testing the prompts using only the programs in each IoT and extended reality (XR) device. By conducting multiple, flexible trials on the cloud, Smart Light was able to complete the same process in about three days—an improvement in efficiency by about 50%. This was a simple on/off use case, but the benefits will be even more substantial in more complex cases.
Smart Light (link resides outside of ibm.com) is a Tokyo-based company founded in 2011. With a team of four employees, Smart Light specializes in integrating lighting, air conditioning and other electricity-powered systems in office buildings and commercial facilities. Leveraging open systems and combining them with cutting-edge IoT and AI technologies, the company focuses on creating energy-efficient and value-generating solutions.
1ft-seabass (link resides outside of ibm.com) is an independent technology studio started by Seigo Tanaka in 2004. Based on web front-end technology, the studio values the perspective of an approach that combines information and interfaces. Recently, 1ft-seabass has been working on development, technical advisement and lectures incorporating technologies such as IoT, XR and gen AI.
Legal
© Copyright IBM Corporation 2025. IBM, the IBM logo, IBM Cloud, and watsonx.ai are trademarks of IBM Corp., registered in many jurisdictions worldwide.
Examples presented as illustrative only. Actual results will vary based on client configurations and conditions and, therefore, generally expected results cannot be provided.