Agentic AI use cases Engineering Managers actually need
Here’s how to make AI agents work for your team this week.
👋 Hey, it’s Stephane. Welcome to my weekly newsletter where I share lessons, and stories from my journey to help you lead with confidence as an Engineering Manager.
Maybe a lot of what you’ve read about AI agents sounds like sci-fi. Cute demos. Twitter hype. Or a cool hackathon project.
But you’re running a team. You don’t care about the hype. You care about solving the same 5 annoying things your engineers complain about every week.
I believe Agentic AI can help you with that. Not tomorrow. Right now.
I’ll show you exactly how.
Get hired as an Engineering Manager (sponsored)
If you are searching for an Engineering Manager job in this horrendous job market and want to make your life easier, I’ve got something for you!
I have created material that simplifies everything:
CV & LinkedIn for Engineering Managers
How to apply (and actually get interviews)
How to prep for behavioural/leadership/people & values interviews
Many people have used the exact same resources in the past year and saw immediate success!
I created this system for myself first, and since I have found so much value from it, I needed to share it with you too!
If you have any questions on it, feel free to reach out at stephane@em-tools.io
Back to this week’s thought.
What is Agentic AI?
Here's the simple version.
Agentic AI refers to artificial intelligence systems that get context and can autonomously make decisions and take actions to achieve objectives without necessary human intervention.
Why should Engineering Managers care?
Agentic AI can help you:
Cut some busywork out of your team’s day.
Make smarter decisions based on data you already have.
Notify you about certain issues before they blow up.
It’s just a matter of thinking and identifying how you can put them to practice and find use cases your team will benefit from.
Here are some ideas:
1. Automate internal operations
You can create agents for things like:
Triage of support tickets: An agent reads, tags, and routes incoming tickets based on past patterns and team ownership.
Prioritise bugs: A bug is raised with some information on it. Then an agent checks the work your team is doing this sprint, your goals, and assesses if the bug should be prioritised now or later.
Release notes from PRs: One agent summarizes merged PRs, another organizes them into user-facing notes. If you chain these agents you could have another one publish them if you’re comfortable, so no one needs to worry about release notes anymore.
2. Better pre-planning with simulations
Planning is hard. Planning across 4 squads is harder. Especially when tech debt needs to be factored in and prioritised against product features.
Try this:
Feed team inputs into a group of agents.
Have one agent play the “tech debt advocate” one act as “PM deadline pusher” and one as “risk analyst”
Have them “debate” and show you different trade-offs: delivery time, risk exposure, and dev morale for different approaches.
This can get you some insight without 9 meetings and 4 whiteboarding sessions.
Try it with your data from Jira, Confluence and GitHub. It will probably find stuff you will have missed.
Note: This should be used as just another tool to help you with decision-making. Don’t fully rely on it to make your decisions. You will find that there is always nuance that your agents might not be able to grasp.
Get 50 Notion templates & The EM’s Field Guide when you subscribe
3. More useful feature design
You’re working with your Product Manager on a new feature and discussing what should be in and out of scope. You can simulate user behaviour with agents to inform your decisions.
Create an agent for your user personas. For instance:
Power users
First-time users
Trolls
Buyers vs. sellers in a marketplace
Simulate interactions with the new feature and get insight on what they like/dislike.
Note: This is best done with real users, but this could help if you don’t have time or a way to be able to talk to the specific user personas you’re building for.
Be careful though
There are real risks. And if you mess this up, you’ll lose your team’s trust fast.
Watch out for:
Data leakage: Agents reading the wrong things (e.g., salary data) if you don’t set up access controls.
Agent chaos: Without rules, things can go off the rails. You need clear roles, limits, and review steps.
Human fear: People worry they’ll be replaced. You must explain how this helps the team and take your team members on the journey with you.
As always: start small & scale what works
Start with one tiny problem and solve it with one agent.
Some easy ideas to start with:
Auto-generate sprint summaries.
Draft release notes based on merged PRs.
Use cases will grow fast once people start seeing results. But they have to see them first.
That’s all, folks!
See you in the next one,
~ Stephane
PS. If you’re taking this management thing seriously, here are more ways I can help:
Follow me on LinkedIn for insights during the week (free).
As mentioned in this article, if you’re looking for an Engineering Manager job, check out this product that I made for you!
Check the products that I offer with a paid subscription - 50 Notion templates, guides, and more. It’s a lot!
Work with me through 1:1 coaching. I find it extremely rewarding to give personalized guidance to engineering leaders who are interviewing for EM roles or are looking to grow in this area.
If you’re a Software Engineer I have a partnership with CodeCrafters – build Redis, Git & more from scratch (40% off with my link).
If you're preparing for interviews, I am also writing this newsletter for architecture design that you might be interested in:
