AI and Developers

AI is influencing software development in two ways: building software using AI and building AI into software.

The first involves using AI tools like GitHub Copilot to reduce developers’ cognitive load, decrease keystrokes, and provide learning assistance.

But the focus here is on the second case: about integrating AI into software to enhance its capabilities and make applications smarter.

Intelligence is the new baseline for modern apps

Intelligence is now the new baseline for modern apps. A few years ago, discussions around modern applications revolved around DevOps, cloud-native development, resilient code, APIs, integration, and reducing technical debt.

Today, the conversation has shifted to infusing AI into applications to make them smarter. Responsible AI is critical—it builds trust, prevents harm, and ensures technology benefits everyone.

But developers are key. AI services are smart, but on their own, they’re not enough. Developers are needed to build the business logic around these AI services to achieve goals and create agentic applications.

Intelligent App - High Level

The high-level architecture of an intelligent app consists of three tiers: the app itself, the front-end client, and the back-end services.

The front-end could be a user interface or another system, which in Microsoft talk is often called a Copilot — essentially, Copilot is the UI for AI. The back-end services can be categorized into AI and non-AI services. A non-AI service might include a weather API or a calendar API, while AI services enable intelligent functionality.

Now, let’s drill down into those different AI services.