The parts of my job I’d happily hand over to a machine
Reflecting on human-centric versus AI-centric work.
For more than seven years, I was a product manager at a software company. Product managers guide the direction of the product based on feedback from customers, sales, and leadership teams. They keep the company updated on the progress of projects and initiatives. And they constantly juggle tasks and deadlines when projects go off track or something unexpected comes up.
It often felt like herding confused cats out of a building that’s always on fire.
A significant amount of my time was spent buried in documentation. Preparing reports for the leadership team, reviewing spec documents for new features (or writing them myself), and preparing release documentation for customers that explained new features and bug fixes.
I look back at the amount of time I spent on much of this work and how it limited what I could get done during the day. Today, that role would be incredibly different, thanks to AI.
David Pierce, Editor-at-Large at The Verge, said in an interview:
The thing that I am hopeful for, and frankly already grateful for in the technology, is that it can do a lot of the busywork I don’t want to do. It’s not creative. It’s not even thoughtful. It’s just administrative.
We can all find better things to do than turning an Excel file into bar graphs in a PowerPoint. So if we can automate that out of existence, I’m actually extremely hopeful for what we can all go find to do instead.
Product management is a critically important role in many companies, but also one that is drowning in tasks. The sources of data are scattered, like customer data, the upcoming project plans, a calendar of due dates, analytics, and more. It’s a chore to pull everything together.
What I would hand over to AI
The same job, today, would be almost unrecognizable. I’d have more time for the things that make a product manager role interesting (thinking about the direction of the product) and spend less time on “paperwork.”
For example:
AI could spot patterns. I used to pore through hundreds of customer support tickets, looking for things that customers were complaining about or asking for. Those with the most requests often received higher priority, but identifying them was an intensely manual process.
AI could create a design prototype. For new product features, developers would work from a mock-up that was painful to create. AI could create this in seconds.
AI could write reports. I produced a monthly report on each project’s status for the leadership team, but the data lived in multiple systems. On top of that, the CEO liked to see the report in a very specific format. All of that could be handed over to AI. I could have even produced more frequent reports with the time saved.
AI could write release notes. AI could determine from the project management and coding tools which features and bug fixes actually made it into a release and draft the release notes into a format suitable for customers. This process used to take hours (if not days) to compile.
What I could have done with the extra time
At a guess, I think the current version of AI tools could have saved me at least 10 hours per week. Maybe more, during the weeks when I was creating design prototypes or writing release notes.
And none of these required human creativity.
Spotting patterns is simply compiling data. The judgment for “What should we build?” is still human.
Design prototypes were based on well-established product specifications, just applied to a new feature (like “This button always goes in the upper-right corner, and when you click it, it does XYZ thing”). Developers simply needed a visual reference so they could code the new feature.
Reporting was also mainly compiling data, with the addition of formatting. Explaining the project’s progress to the leadership team would still come from a human.
Release notes were taking a description of a bug fix written in “developer speak” and translating it into something a customer could understand.
Without the hours spent on these tedious (but necessary) tasks, I could have:
Talked to customers. Product managers at bigger companies with larger teams often talk to customers to understand their pain points. As the sole product manager, I rarely had time for this and had to rely on support tickets written by the customer service team.
Made better product decisions. Since my process of looking through support tickets was so manual, customer needs and pain points were often missed.
Fixed things myself. The company had hundreds of outstanding bugs to fix at any given time. Some were incredibly minor, so they were never prioritized. But with vibe coding, it’s likely that I could have learned to apply a fix myself and let a developer review it. Fewer outstanding bugs, happier customers.
I’ve always tried to automate repetitive tasks throughout my career. I did as much as I could back then, but there were simply limits to what I could do — especially with some of the most time-consuming tasks. Generative AI has changed that.
Of course, I recognize that this also blurs the lines for different roles. Should a product manager be expected to code bug fixes? Maybe not (though I would have enjoyed just fixing something rather than letting it linger).
My point is that I would have found the job much more fulfilling if I had been able to spend more time on the work that required human judgment. Talking to customers would have been incredibly valuable, but it wasn’t a requirement to move forward. So it got pushed to the bottom of the list. Creating a report for the CEO — while tedious — was an expectation of the job.
Thinking about a career change? Download my guide: 5 Types of Career Pivots.
If you want to support my work as a writer, you can subscribe to receive additional issues I publish.
Have a work story you’d like to share? Please reach out using this form. I can retell your story while protecting your identity, share a guest post, or conduct an interview.