Selected work
Complex domains.
Practical engineering.
Examples of how AI engineering and software architecture come together in industrial and business applications.
AI-assisted troubleshooting for industrial service operations
Role: Tech Lead & AI Engineer
Three-person engineering team
The challenge
Industrial troubleshooting requires people to combine technical documentation, historical cases, operational data, and specialist knowledge. Finding and connecting that information can be difficult and time-consuming.
The contribution
Paul contributed technical leadership and hands-on AI engineering to an industrial innovation initiative exploring AI-assisted troubleshooting. The approach combined knowledge retrieval, specialist agents, and operational context to help users investigate issues and identify relevant next steps.
The intended value
Better preparation, more consistent guidance, and more efficient troubleshooting, with human expertise and judgement central to the approach.
This example is anonymised. It describes innovation and proof-of-concept work; quantified operational results and production adoption are not claimed.
An AI-native insurance broker platform
AI agents, conversational interfaces, and event-driven workflows at the core of the solution.
Communication triggers and workflow configuration are planned capabilities.
Current development
AGIS is an AI-native insurance broker solution under development. Alongside the conventional user interface, AI agents and chat interfaces use tools exposed through the Model Context Protocol (MCP) to work with the application’s data and business capabilities.
Access extends across the data layer, subject to permissions and defined exceptions. This enables AI to interact with the business application through its tools and services.
An implemented event-driven workflow generates insurance suggestions as proposals for broker review.
Planned capabilities
The planned product expands this approach to configurable agent workflows triggered by database activity, customer communications such as email and phone interactions, and scheduled events.
Planned configuration will determine which workflows run, under whose authority, and how their outputs enter the business process.
A foundation for future solutions
The foundation is being developed with potential reuse across other business solutions in mind. Broader reuse remains a development objective.
AGIS is under development. Planned communication triggers and configuration capabilities are not presented as available features.
Let’s find a useful starting point
What would you like to improve or build?
Tell us about the challenge, your existing systems, and what success would look like.