Three months ago, I was in a meeting at our office in Islamabad, taking notes by hand while half-listening to someone explain a deployment issue. My developer's instinct kicked in: this is inefficient. But I wasn't thinking about the problem, I was thinking about my own cognitive load. I had a 1:1 with a junior dev afterwards where I needed to give difficult feedback, then a stakeholder call where I had to translate technical chaos into business language. By evening, I hadn't processed a single email properly. That's when I realized: I'm managing like it's still 2010, with AI-era problems.
The article I read this week hit something raw because it wasn't about whether AI is useful, it was about the specific soft skill gaps that AI actually fills. And I've been thinking about those gaps all wrong.
The Meeting Problem (That's Actually a Coordination Problem)
Before I became a manager, I thought meetings were just about status updates. Now I understand they're about creating shared understanding and capturing decisions. The problem? I've been treating my meeting notes like documentation when they're really accountability records.
When I switched to AI-assisted meeting transcription with actual structure, something shifted. The AI doesn't just transcribe, it extracts who owns what, what the decision was, and why we made it. No more "action items" buried in paragraph seven of rambling notes. No more stakeholders claiming they never heard about a commitment.
What matters though: I still review everything. I've seen the AI interpret ambiguous comments wrong. That human-in-the-loop piece isn't optional, it's the entire system working.
Managing Up Is a Communication Problem, Not a Thinking Problem
This one stung because I've wasted entire weekends preparing reports that nobody read. I'd include technical details thinking that more information equals better decisions. What I actually needed was translation, not elaboration.
The realization is uncomfortable: I was creating noise instead of clarity. AI helping me translate deployment metrics into revenue impact, or infrastructure decisions into business risk, that's not dumbing down the information. That's finally speaking the language my stakeholders actually understand.
But here's where I'd push back on the article slightly: the AI isn't your sparring partner for practicing difficult conversations. It's a feedback tool, not a replacement for actually developing those skills. I use it to review my tone before a performance conversation, sure. But if I'm relying on AI to know what to say when someone's about to quit, I've failed as a manager.
The Inbox Chaos Reveals a Deeper Issue
I keep everything in email because it feels like a record. That's lawyer thinking, not builder thinking. What I actually need is triage: what requires my immediate attention, what needs delegation, what can wait.
Using AI to summarize threads and draft replies for routine stuff, I'm honestly still skeptical about this one. Here's my friction point: every message with my signature carries my judgment. If I'm auto-replying to 50 emails daily with an AI agent, I'm not managing communication. I'm creating plausible deniability.
The governance angle the article mentions is crucial. I'm not uploading confidential client data to random AI tools. That's not caution, that's basic professional responsibility.
What I'm Actually Doing Differently Now
I've set up:
- Meeting recording + structured transcription (Otter.ai for now, with manual review)
- Executive summary generation for stakeholder updates (Claude prompt engineering)
- Scheduled one-on-ones with templates that remind me of context before each conversation
What I haven't delegated to AI: the actual difficult conversations, the feedback decisions, or the relationship-building work. That's the stuff that actually makes me a manager worth having.
The Uncomfortable Truth
The article's title claims managers who don't use AI have "already lost." I think that's hyperbole. What's actually true: managers who don't automate their cognitive overhead are wasting capacity on low-value work.
That capacity should go toward listening, coaching, and making judgment calls under uncertainty, the things only humans can do well.
Your Turn
Are you using AI to amplify your thinking or to avoid developing soft skills? Because those aren't the same thing. I'd genuinely like to know: what's one soft skill area where you've tried AI assistance and felt like it actually worked?
Source: This post was inspired by "The Manager Who Doesn't Use AI Has Already Lost: Soft Skills" by Dev.to. Read the original article