Create

Sign in to ReadmeX

or

How we handle your data: Privacy Policy

ReadmeX

Your AI reading companion

What would you like to catch up on?

Pick a briefing, or ask your own question.

Your community and people briefings stay personal to you.

Story

MUA Combines AgentRun, ScriptRun and ChainRun

AI summary

An article on MUA’s hybrid workflow describes three device-automation execution modes: AgentRun makes decisions based on the current page, ScriptRun replays verified fixed operations without calling a large model, and ChainRun combines script and agent nodes in a serial workflow. In a comparison on one cloud phone, both modes completed a three-day Beijing weather inspection. The ChainRun version reduced model usage by reusing navigation as a script, while total task time was broadly unchanged. The article presents the comparison as a newly run test in the example and does not identify it as an independent evaluation.

In-depth insight

The workflow lets teams reuse verified navigation while retaining agent judgment for changing page content. The example also distinguishes lower model usage from end-to-end speed, because the script’s runtime and checks were included in the total ChainRun time.

每天打开同一个 App,进入同一个入口,再面对不断变化的数据、页面和业务条件——这是一类很常见的企业手机任务。

上一期,我们介绍了 OS Agent 如何替企业操作手机和电脑,并用商品调研展示了从任务描述到结果交付的过程。这一期从最常用的 AgentRun 出发,介绍用户如何把固定流程沉淀为 ScriptRun,再通过 ChainRun 组合两类能力。

让跑通的固定步骤可以复用,让变化的内容仍能在线判断。 这听起来有些抽象,不妨先落到一个具体任务上——它无需登录,却定期需要重复一遍。

假如你要做这样一个任务:“每天在云手机上打开浏览器,进入固定的天气预报页面;读取北京未来三天的预报,把日期与天气、最高/最低温和风力逐一配对;如果昼夜温差明显,就提醒增减衣物、给出出行建议;页面上没有的字段要明确标注;最后,把结果整理成一份当天的天气巡检结果。”

#MUA#OS Agent#AgentRun#ScriptRun#ChainRun#Skill#Via Browser#China Weather Network

Original · 9 min readRead the original
Group chat

No comments yet. Start the conversation.