Asana借助Codex在两周内完成了五年的工程工作

大约两周内，Asana 完成了原本预计需要五年才能完成的工作。

Asana 帮助团队协调工作。其平台利用 AI 代理和自动化功能，帮助客户管理、跟踪并执行这些工作。Asana 将同样的方法应用于其工程组织，在那里，人们使用由前沿模型驱动的 OpenAI Codex 来处理大型代码库变更，然后审查并批准每项提议的更改。

对于这个项目，Codex 帮助 Asana 的工程师移除了 Enzyme——一个过时的测试系统，它曾使公司的代码难以升级。模型和基础设施成本约为 12,000 美元，而相比之下，Asana 之前的人员配置计划估计需要约 600 万美元。

“并非每个长达数年的项目都会压缩到几周内完成。但代理可以为工程师提供更多专注于工艺的空间——并让曾经不可能的工作值得尝试。”

——Amritansh Raghav，Asana 首席技术官

他们旧的测试工具 Enzyme 已不再积极维护，并成为 Asana 前端栈现代化的阻碍。Asana 使用由前沿模型驱动的 Codex 来完成这项工作。从一个五句话的提示开始，最多四个编码代理并行工作，每个代理在代码库的独立副本中操作。一名工程师每天检查两次进度，并审查每项提议的更改。更简单的指令比更复杂的设置效果更好。

经过跨越两个日历周的 1.5 周工程努力，Enzyme 被完全移除。模型和基础设施成本总计约 12,000 美元。相比之下：之前的计划预计至少需要五年，估计成本约为 600 万美元。这一经历改变了公司认为哪些长期运行的软件项目是切实可行的。

随着这次迁移的完成，Asana 可以在其他迁移、重写和性能问题上测试代理，而这些任务曾被认为需要数年时间。团队希望这将为工程师提供更多专注于工艺的空间，同时人们继续审查工作。

In about two weeks, Asana completed work it expected to take five years.

Asana helps teams coordinate work. Its platform uses AI agents and automations to help customers manage, track, and execute that work. Asana brings the same approach to its engineering organization, where people use OpenAI Codex, powered by frontier models, to tackle large codebase changes, then review and approve each proposed change.

For this project, Codex helped Asana’s engineers remove Enzyme, an outdated testing system that had made the company’s code harder to upgrade. Model and infrastructure costs came to about $12,000, compared with Asana’s roughly $6 million estimate for the previous staffing plan.

“Not every years-long project will collapse into weeks. But agents can give engineers more room for craft—and make once-impossible work worth attempting.”

—Amritansh Raghav, Chief Technology Officer, Asana

Their old testing tool, Enzyme, had fallen out of active maintenance and was becoming a blocker to modernizing Asana’s frontend stack. Asana used Codex, powered by frontier models, to do the work. From a five-sentence prompt, up to four coding agents worked in parallel, each in a separate copy of the codebase. An engineer checked progress twice a day and reviewed every proposed change. Simpler instructions worked better than a more elaborate setup.

After 1.5 weeks of engineering effort spread across two calendar weeks, Enzyme was fully removed. Model and infrastructure costs totaled about $12K. For comparison: the previous plan was expected to take at least five years and estimated to cost roughly $6M. The experience changed which long-running software projects the company believes are practical to take on.

With this migration complete, Asana can test agents on other migrations, rewrites, and performance problems it once assumed would take years. The team hopes this will give engineers more room to focus on craft while people continue to review the work.