TE
科技回声
首页24小时热榜最新最佳问答展示工作
GitHubTwitter
首页

科技回声

基于 Next.js 构建的科技新闻平台,提供全球科技新闻和讨论内容。

GitHubTwitter

首页

首页最新最佳问答展示工作

资源链接

HackerNews API原版 HackerNewsNext.js

© 2025 科技回声. 版权所有。

Ask HN: What's the hardest part of building AI Agents?

2 点作者 kilroy1235 个月前
I know I'm struggling. Am I the only one who is confused and struggling to build AI Agents? What are you having problems with?

1 comment

hypoxia5 个月前
My $0.02: it&#x27;s too hard to build and iterate on complex workflows.<p>Every agent uses a meta-workflow (eg. ReAct is plan-&gt;act-&gt;observe, with some added steps to check for completion etc.).<p>The teams that have been successful with agents do so by building better but more complex workflows.<p>Most notably, AlphaCodium&#x27;s &quot;From Prompt Engineering to Flow Engineering&quot; <a href="https:&#x2F;&#x2F;github.com&#x2F;Codium-ai&#x2F;AlphaCodium">https:&#x2F;&#x2F;github.com&#x2F;Codium-ai&#x2F;AlphaCodium</a><p>Our current tools don&#x27;t do a great job of making it simple to build and iterate on these workflows.<p>For example, here&#x27;s a HN post from yesterday where a user created their own workflow management platform because of their frustration with the leading tooling providers: <a href="https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=42299098">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=42299098</a><p>I think once we get this tooling right and start to build more expertise in the process of flow engineering, we&#x27;ll start to faster improvement in agent quality.
评论 #42325092 未加载