AI tasks everyone knows about
TL;DR: I use AI mostly for proofreading, meeting notes, and mockups. Together, they save me less than 5% of time/week. Not game-changing, but a decent start.
These days, I often kick off conversations with colleagues and friends by asking how they use AI. It’s become a great conversation opener!
The answers are usually predictable: text editing, quick Q&A chats, summarizing meeting notes. All useful, but none of these have taken my productivity to the next level. Still, let’s break them down for accuracy and order.
For context: I’m a product manager, work ~50 hours a week, not a native English speaker, don’t code, and spend a lot of time writing docs. Here are my three most common use cases of using AI at work:
The image is by Harper Sunday from Unsplash
1. Proofread and enhance my texts: +1% time savings
Honestly, it came as a surprise for me when I started calculating how much AI saves me here - not much! My general approach to using AI for texts is as follows:
Dump the ore. I once read that words are like ore: you have to mine them before you can extract gems (good text). But first, you have to mine the ore - meaning, write something, anything. So my first step is to write everything in my head - roughly, awkwardly, without elegant metaphors.
Shape the structure. Then I cut down, rearrange and add more structure to a document. Remove unnecessary words and sentences. Anything interesting but not essential goes into the Appendix.
Make it convincing. This is where AI comes in. I copy-paste my draft into Claude (my current favorite) and add a prompt with (a) main purpose of the doc, (b) who is the target audience and (c) request to proofread and enhance the text.
Polish a final version. When AI is ready with the final text, I copy-paste it to my original doc and compare it with the original text - which might explain why it doesn’t feel like a huge time saver! So the final text is always a mix of my own text and AI-generated bits.
What I use: I use corporate versions of LLMs: you know that if you put it in public chatGPT or whatever, they might use it for training purposes, right? I tried Gemini, chatGPT and Claude. Claude works best for me.
Time savings: 1% time/week.
2. AI-generated meeting notes: <1% time savings
For full disclosure: I still prefer writing meeting notes myself and apply a rule of working with them as with usual text (see pt.1). The reason behind this is that I usually send meeting notes to either share an important context with the group, or influence group opinion, or highlight critical updates and blockers in a project. AI is terrible at doing all of that. The only time I lean on AI notes is for purely transactional meetings, when I just need to show stakeholders that the meeting happened and action items are clear.
What I use: Zoom’s AI meeting notes.
Time savings: less than 1% time/week.
(Teaser: I did discover a neat “cheat code” for UXR notes - will cover that in the next post.)
3. Generate mockups: +2% time savings
This one’s actually useful. In the pre-AI era, I’d either write PRDs (Product Requirements Docs) without visuals or spend hours hacking in Figma to create rough mockups. Now, with tools like Loveable or v0+Vercel, it takes me a couple minutes to write a detailed prompt and iterate on the design until I get a "good enough” version I can use in my PRD. This is the most helpful use case so far. The detailed algorithm of working with it is as follows:
Create a PRD with a UX description
Take a screenshot of existing tool or design system components for style reference.
Upload both to the tool, ask for a design output.
Tweak to a “good enough” version.
It works really great, and I’m glad that design/prototype tools became more affordable with little to no entry barrier. Still, since I don’t write PRDs every week, the net time savings stay modest.
What I use: Loveable (my favorite so far), v0+Vercel.
Time savings: ~2% of my week.
Wrapping up
All in, these basic AI use cases save me around 5% of my weekly time. Not bad for the start, but I’m experimenting with more advanced workflows to see if I can unlock bigger productivity gains.