Replit 推出 Missions:给 Agent 设定目标、预算与独立评审
Replit 给自家 Agent 加了个新玩法:你定目标和预算,它自己跑,还有个 AI 评委按你的标准验收。跟 Cursor、Devin 那类只能下指令的工具比,可控性明显不一样。经常让 Agent 无人值守跑任务的开发者可以看看直播演示。
Replit 宣布推出 Replit Missions,这是其 Agent 产品线的功能更新。Replit 在 2024 年 9 月发布首个普遍可用的 Agent,到 2025 年 9 月,该 Agent 已能自主运行超过 200 分钟。Missions 在此基础上让用户为 Agent 设定一个明确目标,并自行配置预算和边界条件。
运行过程中,用户可以看到 Agent 的可见进度展示,同时系统引入一个独立的 AI 评审机制,按用户预先定义的标准检查 Agent 产出是否达标。Replit 还宣布将于太平洋时间上午 9 点/美东时间中午 12 点举办线上直播,由 samuel_spitz、Ibrahim Sheikh 和 Tulika Gupta 三位团队成员演示该功能。
这类“目标+预算+自动评审”的设计思路,与 Cursor、Devin 等编程智能体产品在可控性上的探索方向一致,对需要在无人值守场景下运行 Agent 的开发团队有直接参考价值。
Replit has always pushed the frontier of what AI can do. In September 2024, we released the first generally available Agent. By September 2025, that same Agent could work autonomously for over 200 minutes. Today, we’re pushing that frontier further with Replit Missions. Replit Missions gives agents a goal to work toward, with budgets and boundaries you set, visible progress, and an independent AI judge checking results against your criteria. Join us at 9am PT / 12pm ET tomorrow for a live webinar with Replit experts @samuel_spitz , Ibrahim Sheikh and Tulika Gupta 🔗 replit.com/discover/put-a… e📊 4 ⚡
This report is categorized under AITOP's "Products" desk, focusing on US frontier AI labs and Silicon Valley technology leaders. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows and US-China foundation model benchmarks and commercialization comparison. Key entities and technical topics under continuous tracking include Replit, agents, coding assistants.
Source attribution: curated from Replit (@Replit).