For engineers at Applied Systems, building AI solutions isn’t just about moving fast — it’s about creating products people can depend on.
That’s why Associate AI Engineer Paul McSlarrow and his peers at the cloud-based insurance provider lean heavily on test-driven development when integrating AI into their product environments.
“By defining expected outcomes first, we can guide the model toward correct behavior, validate requirements earlier and identify edge cases before they become larger issues,” McSlarrow said.
Embracing a responsible approach to AI implementation, McSlarrow has tackled a variety of impactful projects at Applied Systems. For instance, he’s currently working on a new “Send Email” capability that allows users to build email communications directly into an automated workflow. McSlarrow said that the solution is designed to make it easier for customers to create personalized, data-driven communications while reducing the friction that can come with managing complex workflow configurations.
For McSlarrow, AI isn’t just an innovation accelerator — it’s also a catalyst for career growth.
“As someone still early in my career, being trusted with a project that spans so many parts of the stack has been an incredible opportunity to learn, contribute and see how product ideas become customer-facing capabilities,” McSlarrow said.
Below, McSlarrow shares more about how his team leverages AI while managing risk, and the impact of the project he’s currently working on.
Applied Systems is a global provider of cloud-based insurance software and agency management systems.
Where does AI show up in your stack?
AI shows up across our development process, from helping us navigate complex codebases and document our work to supporting testing and accelerating how we move from an idea to a working solution. My own work spans front-end development, APIs, data and cloud-based services, so the value is much broader than generating code. AI can help connect context across those layers, while the engineer still owns the architecture, validation and decisions that determine whether the final product works for the customer.
How do you manage risk?
Speed matters, but so does building something people can depend on. One practice that has been especially valuable is test-driven development. By defining expected outcomes first, we can guide the model toward correct behavior, validate requirements earlier and identify edge cases before they become larger issues. We also review AI-assisted work, test the complete workflow and build safeguards directly into the product experience.
On my current project, we validate required information before a workflow can be scheduled and verify that the selected dataset can support the intended action. We also use feature flags as capabilities evolve and implement controls that limit what AI services can access, including restricting the tools and file types available to the model. AI helps us move faster, but responsibility for accuracy, security and reliability remains with the engineering team.
“AI helps us move faster, but responsibility for accuracy, security and reliability remains with the engineering team.”
What’s an exciting project you’re on?
I’ve been working on a new “Send Email” capability within Atlas data-driven workflows. The feature allows users to build email communications directly into an automated workflow by connecting a dataset, selecting approved templates and recipients from Epic, which is Applied’s agency management system, adding custom content, and using merge variables populated from their data.
One aspect I’m particularly proud of is the amount of validation built into the experience. The system verifies that required fields exist in the selected dataset, surfaces guidance when information is missing and supports users across multiple locales. The goal is to make it easier for customers to create personalized, data-driven communications while reducing the friction that can come with managing complex workflow configurations.
I owned the feature end to end, from concept planning and ticket creation to cross-team coordination, UI development, back-end API work and workflow logic. As someone still early in my career, being trusted with a project that spans so many parts of the stack has been an incredible opportunity to learn, contribute and see how product ideas become customer-facing capabilities.
