How AI Agents Transform the Day-to-Day Life of an Engineering Manager


I know there are plenty of posts and materials about using GenAI in software development, especially how it can act as a junior software engineer working alongside you. However, I haven’t found enough examples of how GenAI can assist managers who lead software development teams—a.k.a Engineering Managers.

This post is a quick summary of how I’ve been using GenAI in my day-to-day work to stay ahead of the game. Please share any additional use cases you’ve found helpful for engineering managers in the comments!

Early Adoption of GenAI at Amazon

In late 2023 and early 2024, Amazon developed numerous internal tools, primarily built on Anthropic Claude models, to assist with daily work. It started with simple ChatGPT-like utilities for correcting grammar and rephrasing and summarizing text based on prompts. I realized it was no longer acceptable to make silly grammatical errors in written communication (whether in emails or Slack messages) when tools like these were readily available.

My first use of GenAI was adopting these text validators. As a non-native English speaker facing challenges in communication, I found these tools immensely improved my ability to clearly convey messages in a less ambiguous way.

A Game-Changing GenAI Agentic CLI

In mid-2024, one of my sister teams developed a GenAI CLI using agentic Anthropic Cluade models. This was a game-changer for me, as it leveraged multiple internal resources—such as wikis, websites, and code repositories—to automate tasks across the software development lifecycle. I became an early adopter of this utility.

As an engineering manager, 70% of my time is spent in meetings with my team and various stakeholders. This often leaves me with limited time to deeply engage in my team’s daily work. However, it’s crucial for me to have a clear understanding of who is doing what—without micromanaging.

Using our custom-built GenAI CLI, I can:

  1. Summarize code reviews from my team.
  2. Dive deeper into specific decisions made during code reviews.

This tool has saved me roughly 50% of the time I’d otherwise spend searching through references to understand the reasoning behind certain choices. Beyond the time saved, it has also enhanced my team’s trust in me as a manager who understands the technology and architecture of the products we own.

Staying on Top of Technical Designs

My team often works on technical architecture and designs for various solutions. While I try to participate in every design review, scheduling conflicts occasionally prevent me from attending. To stay informed, I use Amazon’s internal AI assistants to summarize technical design documents for me.

When architecture diagrams created in tools like Draw.io aren’t immediately clear, I ask the GenAI CLI to explain them. 

Monitoring System Operations

I own the systems my team builds, from inception to operations and eventual deprecation. To stay on top of daily operations, I rely on observability dashboards for a high-level overview. However, for a deeper understanding of specific issues, I use GenAI agents to query and analyze CloudWatch logs, providing me with concise summaries of any problems.

Assisting in Performance Evaluation Cycles

While the role of an engineering manager involves many responsibilities—such as driving technical direction, stakeholder management, and product roadmaps—I believe people management and career growth are the most critical. Assessing my team’s performance incorrectly can have disastrous consequences for their careers. Performance management is something I approach with great care.

Since adopting GenAI tools, I’ve found them invaluable in performance management tasks:

  1. Writing Promotion Narratives: I use our custom-built GenAI CLI to input a draft of a promotion narrative and provide prompts to ensure it aligns with the expectations of the next-level role. Based on the feedback, I refine the narrative accordingly.
  2. Summarizing Individual Contributions: The GenAI CLI helps me gather data from multiple sources—operational tickets, code repositories, design documents, and sprint boards—to build a holistic view of each engineer’s contributions.

Using GenAI tools for these tasks has saved me at least two weeks of effort when preparing performance reviews for a team of 10+ engineers.


Comments

Post a Comment

Popular posts from this blog

WSO2 Stratos - Introducing WSO2 middleware Platform as a Service (PaaS)

Image
Yesterday we released our newest integrated cloud middleware platform, WSO2 Stratos . It is now online and ready for use! For the past few years, WSO2 has been doing many releases based on revolutionary Carbon platform , which helped to consistently improve the platform with introducing latest state-of-art technologies. The initial releases of Carbon product platform intended to improve the features of the individual products. For example, a lot of core features were introduced to WSO2 ESB, WSAS, Governance Registry etc.. Then we focused on ease of integration and more component based model. We introduced Equinox based provisioning model to build your SOA platform by picking and choosing the components as and when you need. Then, the core product platform has been enhanced to support multi-tenancy. With the multi-tenant supported architecture at hand, our brilliant development team was able to introduce the first ever comprehensive middleware PaaS. Why is this so important? Now, WSO2 C...
Read more

Common mistakes to avoid in WSO2 ESB - 1 - "org.apache.axis2.AxisFault: The system cannot infer the transport information from the URL"

In the next few weeks, you could expect a series of blog posts from me which explain the remedies to avoid some common mistakes which we do when working with WSO2 ESB . This is the first of many. During the initial rounds of testing your integration solution or even in production systems, you may have come across the following error. org.apache.axis2.AxisFault: The system cannot infer the transport information from the URL .... Some of you may have struggled hours figuring out a solution to address this issue. Let me explain this error in detail and possible causes of this. So, next time when you see this error in your WSO2 ESB setup, you will not have to spend time unnecessarily googling everywhere. There can be many different causes of this error. I have observed  four main reasons. Cause 1 This is the simplest cause. Suppose, you have a proxy service similar to the following. <proxy xmlns="http://ws.apache.org/ns/synapse" name="MistakesTest...
Read more

Working with HTTP multipart requests in soapUI

Image
You can use HTTP request test step in soapUI to submit messages with various Content-Types.  In this post, we will have a quick look into the multipart/form-data requests in soapUI. multipart/form-data requests usually come into action when you do HTML form submissions with file attachments. For example, have a look at the following HTML form post. <form action="http://localhost:8090/CKFileUploadApp/UploadServlet" method="post" enctype="multipart/form-data"> <input type="file" name="file"> <input type="text" name="paramName"> <input type="submit" name="Submit" value="Upload File"> </form> In the forms like above, we have file upload option as a form input where we can upload a file along with the form submission. In that case, the request's HTTP content type will be multipart/form-data. How can we simulate such a request using soapUI...
Read more