Introducing Courses and Insights on Arm Learning Paths

Introduction

Arm Learning Paths has a straightforward goal: Make it easier for developers to get things done on Arm-based platforms.

For the past few years, two content types have enabled that goal. Install Guides help developers quickly set up the tools and resources they need. Learning Paths provide step-by-step tutorials for accomplishing specific development tasks on Arm. Both are more actionable than a typical blog post and more task-oriented than product docs.

Now, the way developers work is changing. 

AI-assisted development is changing how developers find and use technical knowledge. Developers still visit Arm Learning Paths directly to learn a technology or complete a task. Increasingly, they also ask coding agents for answers, guidance, and code from inside their development environment. In those moments, the quality of the experience depends on what the developer understands. It also depends on whether the agent can find reliable technical context.

Arm Learning Paths needs to support both developers and the AI agents they work with. That is why we are expanding Arm Learning Paths with two new content types: Courses and Insights.

Courses: Build your own expertise

To get the most value from AI-assisted development, developers need enough understanding to ask the right questions, evaluate the answers, and make sound engineering decisions. This is particularly important when working with specialized technologies. Authoritative, domain-specific guidance can help developers build the expertise they need.

Courses are long-form, structured learning experiences that help developers build that foundation. Rather than focusing on one specific task, a Course helps developers build a broader understanding of a technology. They can then apply that knowledge across projects.

That deeper expertise helps developers get more from AI-assisted development. They can judge the output for themselves, identify when something may not be quite right, and recognize when further investigation or expertise is needed.

Each Course is created with subject-matter experts and combines theory with hands-on coding exercises. This helps developers apply their knowledge as they move through the Course. The goal is to take a developer from “What is this?” to “How do I build with this?” Like the rest of Arm Learning Paths, Courses are self-service and free.

Our first Course, Optimize and deploy AI inference with ExecuTorch on Arm, provides approximately eight hours of guided, hands-on learning. It is designed for developers working with on-device AI. Developers learn how to deploy PyTorch models with ExecuTorch on Arm CPUs and NPUs. The Course covers backend delegation, quantization, and performance comparison.

Explore the course and share your feedback through our end-of-course survey. Your feedback will help us improve this Course and future Courses. Visit the Arm Learning Paths site for more Courses in the coming months.

Insights: Give agents better context

The other side of the equation is what your agent knows.

AI agents can still hallucinate, rely on outdated information, or struggle to find authoritative answers about new and specialized technologies. Repeatedly asking an agent to search and interpret the web also consumes time and tokens. There is no guarantee that it will find the right source.

We wanted to create a content type that gives agents a reliable source of answers to highly specific technical questions.

Enter Insights. Each Insight area organizes expert knowledge into a curated set of concise technical questions and answers. Insights provides an authoritative source on learn.arm.com that developers and AI agents can use to answer specific questions.

Developers can also download Insights as structured JSON and use them to provide context directly into their agentic workflows.

At launch, Insights are available for:

• ExecuTorch
• Arm Scalable Vector Extension (SVE)
• Arm Scalable Matrix Extension (SME)

More Insight areas planned over the coming months.

How it all works together

Courses and Insights complement the existing modalities already available on Arm Learning Paths.

Need	Modality	Primary consumer
Learn the technology	Courses	Developer
Get authoritative answers	Insights	Agent and developer
Set up the environment	Install Guides	Developer and agent
Accomplish a task	Learning Paths	Developer and agent

Consider a developer tasked with building an Android application that uses on-device speech recognition with ExecuTorch. The developer has not used ExecuTorch before. They can start with the ExecuTorch Course to understand the framework and build enough expertise to make informed decisions about their application.

Then it is time to build. They use an Install Guide to prepare their development environment, either by following the guide directly or with help from their coding agent. To deploy a Whisper INT8 model or optimize a particular workflow, they can follow the relevant Learning Path directly or with their agent.

Along the way, they encounter the small but important questions that come with real development. Instead of leaving their IDE and searching across multiple sources, the developer can ask their agent. ExecuTorch Insights provides the agent with authoritative, Arm-curated technical context. The agent can then use that information to answer questions directly within the developer's existing workflow.

That is the broader value of Arm Learning Paths. Developers build deeper expertise, agents get better context, and both can get more done on Arm.

Looking ahead

More content is coming across each of these modalities in the months ahead. While Courses and Insights are the next step in expanding Arm Learning Paths, they are not the last.

We are also developing Skills: Reusable capabilities designed specifically for repeatable agentic development tasks. Conversational AI is also coming to Arm Learning Paths, giving developers another way to query and navigate Arm technical knowledge.

Contribute your expertise

Arm Learning Paths is open source and community-driven, with technical and editorial guidance from teams across Arm.

If you have technical knowledge that could benefit Arm developers and their agents , we encourage you to contribute. You can follow this tutorial to author a Learning Path or Install Guide through our GitHub-based PR flow.

Skills are coming soon. If you are interested in contributing a Skill today, contact learn@arm.com.