您是否曾经想暂停自动工作流程以等待人类的决定？

也许您需要在提供云资源，将机器学习模型推广到生产或收取客户的信用卡之前需要批准。

Have you ever wanted to pause an automated workflow to wait for a human decision? Maybe you need approval before provisioning cloud resources, promoting a machine learning model to production, or charging a customer’s credit card.

您是否曾经想暂停自动工作流程以等待人类的决定？也许您需要在提供云资源，将机器学习模型推广到生产或收取客户的信用卡之前需要批准。

In many data science and machine learning workflows, automation gets you 90% of the way — but that critical last step often needs human judgment. Especially in production environments, model retraining, anomaly overrides, or large data movements require careful human review to avoid expensive mistakes.

在许多数据科学和机器学习工作流程中，自动化为您带来了90％的途径 - 但是关键的最后一步通常需要人类的判断。尤其是在生产环境中，模型再培训，异常覆盖或大型数据移动需要仔细的人类审查，以避免昂贵的错误。

In my case, I needed to manually review situations where my system flagged more than 6% of customer data for anomalies — often due to accidental pushes by customers. Before I implemented a proper workflow, this was handled informally: developers would directly update production databases (!) — risky, error-prone, and unscalable.

就我而言，我需要手动查看我的系统标记超过6％的客户数据异常数据的情况 - 通常是由于客户意外推动。在我实施适当的工作流程之前，这是非正式处理的：开发人员将直接更新生产数据库（！） - 风险，容易出错且不计。

To solve this, I built a scalable manual approval system using AWS Step Functions, Slack, Lambda, and SNS — a cloud-native, low-cost architecture that cleanly paused workflows for human approvals without spinning up idle compute.

为了解决这个问题，我使用AWS步骤功能，Slack，Lambda和SNS建立了一个可扩展的手动批准系统，这是一种云本地，低成本的体系结构，可干净地暂停人工批准的工作流，而无需旋转闲置计算。

In this post, I’ll walk you through the full design, the AWS resources involved, and how you can apply it to your critical workflows.

在这篇文章中，我将带您完成完整的设计，所涉及的AWS资源以及如何将其应用于关键工作流程。

Let’s get into it 👇

让我们来👇

The Solution

解决方案

My application is deployed in the AWS ecosystem, so we’ll use Aws Step Functions to build a state machine that:

我的应用程序已部署在AWS生态系统中，因此我们将使用AWS步骤函数来构建一台状态计算机：

Here is a youtube video showing the demo and actual application in action:

这是一个YouTube视频，显示了演示和实际应用程序：

I have also hosted the live demo app here →👉 https://v0-manual-review-app-fwtjca.vercel.app

我还在这里托管了实时演示应用→👉https://v0-manual-review-app-fwtjca.vercel.app

All code is hosted here with the right set of IAM permissions.

所有代码均在此处托管，其中包含正确的IAM权限。

Step-by-Step Implementation

分步实现

The flow above generates a dataset, uploads it to AWS S3 and if a review is required, then invokes the Manual Review lambda. On the manual review step, we’ll use a Task lambda with an invoke on WaitForTaskToken, which pauses execution until resumed. The lambda reads the token this way:

上面的流量生成数据集，将其上传到AWS S3，如果需要进行审查，请调用手动评论lambda。在手动审核步骤中，我们将使用一个任务lambda和WaitfortaskToken的Invoke，该任务停止执行直至恢复。 Lambda以这种方式读取令牌：

This Lambda sends a Slack message that includes the task token so the function knows what execution to resume.

此lambda发送了一个休闲消息，其中包含任务令牌，因此该函数知道要恢复执行什么。

2. Before the we send out the slack notification, we need to

2。在我们发送松弛通知之前，我们需要

I followed the youtube video here for my setup.

我在此处遵循YouTube视频进行设置。

3. Once the above is setup, setup the variables into the web-hook step of the slack workflow:

3。一旦设置了上述，将变量设置为Slack Workflow的Web-Hook步骤：

And use the variables with a helpful note in the following step:

并在以下步骤中使用有用的说明的变量：

The final workflow will look like this:

最终的工作流程看起来像这样：

4. Send a Slack Notification published to an SNS topic (you can alternately use slack-sdk as well) with job parameters. Here is what the message will look like:

4.将发布的Slack通知发送到SNS主题（您也可以使用Slack-SDK）与作业参数一起发送。这是消息的样子：

This Lambda sends a Slack message that includes the task token so the function knows what execution to resume.

此lambda发送了一个休闲消息，其中包含任务令牌，因此该函数知道要恢复执行什么。

5. Once a review notification is received in slack, the user can approve or reject it. The step function goes into a wait state until it receives a user response; however the task part is set to expire in 24 hours, so inactivity will timeout the step function.

5。一旦在Slack收到审核通知后，用户就可以批准或拒绝。步骤功能进入等待状态，直到它收到用户响应为止；但是，任务零件设置为24小时内到期，因此不活动将超时该步骤功能。

Based on whether the user approves or rejects the review request, the rawPath gets set and can be parsed here: code

根据用户是批准还是拒绝审核请求，将设置RAWPATH并可以在此处解析：代码

The receiving API Gateway + Lambda combo:

接收API网关 + lambda组合：

Example:

例子：

Note: Lambda configured with WaitForTaskToken must wait. If you don’t send the token, your workflow just stalls.

注意：用WaitfortaskToken配置的Lambda必须等待。如果您不发送令牌，那么您的工作流程只是失速。

Bonus: If you need email or SMS alerts, use SNS to notify a broader group.Just sns.publish() from within your Lambda or Step Function.

奖励：如果您需要电子邮件或SMS警报，请使用SNS通知更广泛的组。just sns.publish（）从您的lambda或step函数中。

Testing

测试

Once the manual approval system was wired up, it was time to kick the tires. Here’s how I tested it:

手动批准系统连接起来后，就该踢轮胎了。这是我测试的方式：

I tested all major paths:

我测试了所有主要途径：

Behind the scenes, I also verified that :

在幕后，我还验证了：

I highly recommend testing not just happy paths, but also “what if nobody clicks?” and “what if Slack glitches?” — catching these edge cases early saved me headaches later.

我强烈建议您不仅要测试快乐的道路，还建议“如果没人点击怎么办？”和“如果松弛的故障怎么办？” - 尽早抓住这些边缘案件使我头痛。

Lessons Learned

经验教训

Wrapping Up

总结

Adding human-in-the-loop logic doesn’t have to mean duct tape and cron jobs. With Step Functions + Slack, you can build reviewable, traceable, and production-safe approval flows.

添加人类的逻辑并不一定意味着胶带和Cron工作。使用步骤功能 + Slack，您可以构建可审查，可追溯和生产安全的批准流。

If this helped, or you’re trying something similar, drop a note in the comments! Let’s build better workflows.

如果这有所帮助，或者您正在尝试类似的事情，请在评论中放下注释！让我们构建更好的工作流程。

Note: All images in this article were created by the author

注意：本文中的所有图像均由作者创建