flâneur

Best practices for using AI coding Agents

augmentcode.com · 1,506 words · saved by 1 readers

Agents are a fascinating technology. They're so new and powerful that we're still discovering their capabilities and novel ways to use them effectively. In this blog post, we outline best practices for using Augment Agent that we've discovered both internally and with our fantastic group of early testers. To make it easily digestible, we've organized it as a FAQ. Many recent advancements in AI mirror how humans actually work: This is particularly true for Agents. The key to working effectively with them is to treat it as collaboration with your fellow engineer (less experienced, but very smart). Everything else naturally follows from this approach. At Augment, we see engineers start using Agent for virtually everything—from frontend and backend development to reviewing PRs, writing documentation and design docs, and helping with brainstorming. At the same time, it's pretty clear that there are some tasks in which getting value from agents might be much easier than in others. Here are t

Agents are a fascinating technology. They're so new and powerful that we're still discovering their capabilities and novel ways to use them effectively. In this blog post, we outline best practices for using Augment Agent that we've discovered both internally and with our fantastic group of early testers. To make it easily digestible, we've organized it as a FAQ. Table of contents: Preamble ‍ What are Agents good for? ‍ How should I give the Agent a task? ‍ What if the Agent is not doing what I want? ‍ What if I don't trust the Agent? ‍ What should I do while the Agent is running? ‍ How do I r

saved by

related reading