BoilerBot Intelligent Assistant for Gas App Uk

The Gas App Uk app was envisaged by a gas boiler engineer who recognized the power of a mobile app to help other engineers in their day-to-day work. In a continuing mission to use cutting edge technology to increase the app’s utility Gas App Uk’s founder asked Exploding Phone to harness Artificial Intelligence to provide quick and simple access to crucial but sometimes hard to access information.

A new boiler is typically supplied with a set of hardcopy Installation and Maintenance manuals however these documents are not always subsequently available at the boiler’s location. This results in service engineers needing to search the internet for the appropriate manuals and guides. From mid- 2025 GasAppUk users have the option to ask the app’s BoilerBot ‘Assistant’ to quickly retrieve information about gas boilers.

The system employs ‘Chat Flow’ scripts to perform certain actions, for example, outputting pre-determined text, to prompt the user with a set number of common questions, ask for user text or voice input and then to execute AI model queries. This gives BoilerBot great flexibility to configure itself, either by selecting a Boiler manual to run AI model queries against or to configure itself into ‘Support’ mode and answer queries about Gas App Uk itself. By establishing key information within the Chat Flow, for example, make and model of boiler, the pointers (or AI tokens) that need to be sent to the AI platform are minimized thus simplifying and reducing the cost to retrieve the answer. The Chat Flow is bespoke to GasAppUk and separate from the third party AI component of the system.

In addition to designing and building the update to Gas App Uk a vast set of boiler related documents have been uploaded and vectorized. The resulting Vector Stores are multi-dimension representations of chunks of data stored in a hierarchical tree. The AI system then carves “pathways” through this vectorized data to generate output text.

When in ‘Support’ mode, Boiler Bot uses a specially created Q&A document to derive answers to app user’s questions about Gas App Uk, while for Boilers queries the model relies almost entirely on information obtained from the input manuals.

Careful consideration needed to be given to the choice of AI model. In this case the customer required the system to be both sufficiently quick and accurate to ensure users appreciated the benefit of using BoilerBot, while also managing the cost associated with using the AI platform.

As with most Exploding Phone systems the back-end Content Management System (CMS) has a crucial role in ensuring the solution. . In this case the CMS has been extended to allow app administrators to manage the app’s FAQs and add boiler manuals to the system.