The first time I used ChatGPT for work, I remember thinking: "This is it. I'll never write an email again."
That was two years ago. I'm still writing emails.
But here's what changed: my relationship with time itself. Not in the Instagram productivity-guru sense — I'm still slow at mornings, still miss the 7:42 train from Kalyan twice a week, still spend 20 minutes deciding what to eat for lunch. What changed is how I spend the 8 hours I'm actually at my desk at Morningstar.
The productivity myth around AI is that it makes you 10x faster. The truth is messier. It makes you differently fast. It speeds up some things (report writing, data cleaning), creates bottlenecks in others (decision-making, creative thinking), and occasionally just sits there while you stare at a blank screen wondering if what you're asking it to do is even worth doing.
After two years of experimenting with every tool from Claude to Perplexity to a dozen others, I've figured out what actually works. Not productivity theater. Not the stuff that looks good in a LinkedIn post. The real stuff.
The Tools I Actually Use (Not the Ones Everyone Talks About)
Let me start here: I don't use AI for everything. And that's the first honest thing most productivity posts won't tell you.
ChatGPT Plus for Writing and Thinking
I pay ₹2,000 a month for ChatGPT Plus. It's become the most useful ₹24,000-a-year expense at my job.
But not for what you'd think. I don't use it to write my analysis reports — those need my brain, my data, my point of view. Where it helps: I use it as a thought partner. I'll write a messy paragraph about why a mutual fund's expense ratio matters, paste it into ChatGPT, and ask "What's the argument I'm actually trying to make here?" It forces clarity. Sometimes it rewrites it better than I could in 10 minutes of staring. Other times, I realize my original version was actually clearer, and I keep it.
The speed gain? I save about 90 minutes a week on editing. That doesn't sound like 10x. That's because it isn't. It's more like 1.3x. But 90 minutes a week compounds.
Claude for Code and Technical Work
Here's where I actually see a genuine productivity leap.
I'm not a programmer, but my job involves Excel, SQL, and Python scripts to clean financial datasets. Claude is better at this than ChatGPT, full stop. I can paste a dataset, describe what I need in plain English, and get working code 80% of the time. The other 20% needs debugging, but I still save 4-5 hours a week compared to struggling through Stack Overflow or asking colleagues.
That's closer to 3-4x faster. And that time compounds into something real: fewer sleepless Mondays before client presentations.
Perplexity for Research and Current Information
ChatGPT's knowledge cuts off. Perplexity doesn't. For financial analysis, current data matters. When I need to cross-check recent market movements, regulatory changes, or fund performance updates, Perplexity saves me 20 minutes of jumping between ET Markets, Groww, and government websites.
Not revolutionary. Just useful.
Where AI Actually Saves Real Time (and Where It Doesn't)
Here's the cold truth: AI doesn't save time on everything. It saves time on *specific* things. You need to know which ones.
| Task Type | AI Speed Gain | Why It Works (or Doesn't) |
|---|---|---|
| Writing emails, summaries, meeting notes | 2-3x faster | AI handles structure and tone well; you still need to add context and personality |
| Data cleaning and script writing | 3-4x faster | Clear rules, repeated patterns; AI excels at this mechanical work |
| Strategic decisions and analysis | 0.8x (slower) | AI hallucinates confidence; you end up double-checking everything, adding time |
| Learning new concepts | 1.5-2x faster | AI explains well but can oversimplify; you still need deeper reading for mastery |
| Creative writing, original ideas | 1x (no gain) | AI is derivative; the real work is your original thinking |
| Finding and filtering information | 2x faster | AI synthesis saves scanning time, but quality depends on how you ask |
Notice what I didn't include? Tasks that require judgment. That's intentional.
I used to think AI would help me make better investment decisions faster. Spoiler: it doesn't. It gives me 10 confident-sounding opinions, and I end up spending *more* time verifying which ones are actually sound. The decision-making part — the part that actually matters — is still 100% me.
The Real Productivity Hack: Knowing What NOT to Automate
This is where most people get it wrong.
They see AI and think "I can offload this." But productivity isn't about offloading everything. It's about offloading the *right* things so you have mental energy for the things that matter.
The first three months, I tried to use AI for everything. I'd ask it to draft analysis, write client emails, even suggest portfolio recommendations. It was fast. It was also mediocre. My boss noticed. My clients noticed. And I was spending more time cleaning up the output than if I'd just done it myself.
The shift came when I stopped thinking "How can AI do this?" and started thinking "What part of this is soul-draining and repetitive enough that I shouldn't be doing it?"
For me, that's:
- Data formatting: Converting CSV files into readable formats, cleaning columns, removing duplicates. AI + some manual tweaking: 15 minutes. Manual: 45 minutes. That's a no-brainer.
- Email drafts: I write the core message, AI polishes tone. Not vice versa.
- Research summaries: When I need to understand what happened in markets this week, AI aggregates. I verify.
What I never let AI touch:
- Client recommendations (this is where I add actual value)
- Financial logic (I've seen AI confidently recommend terrible decisions)
- Original analysis (if it's just summarizing, I'm not a data analyst, I'm a bot with a salary)
The commute from Kalyan to Mumbai every day used to feel like wasted time. Now I use those 90 minutes to think without a screen. To let ideas percolate. To read actual books instead of blog posts. I'm convinced that hour-and-a-half of thinking time is worth more than an extra 5 hours of AI-assisted busy work.
How to Actually Integrate AI Without Becoming Dependent on It
The trap is real. You start using AI, it feels faster, and suddenly you're asking it for everything. Then you forget how to do things without it. Then you realize you've outsourced your judgment.
Here's what I do to avoid that:
One: The 70% Rule
I only use AI if it can do 70% of the work correctly on the first try. If it's less, I'm spending more time correcting than I'd spend doing it myself. If it's more, I'm not learning anything.
Code? 80% of the time it's good. Writing analysis? 60% of the time. Investment decisions? 20% of the time. I adjust my tool use accordingly.
Two: Audit Every Output
I never use AI output without checking it. Always. Every single time. Even when it looks right. I've caught hallucinations that would've made me look stupid in front of clients.
This takes time. But it's non-negotiable.
Three: Keep One Thing Manual
I still write one analysis report a month completely by hand — no AI assist, no templates, just me and Excel. It keeps me sharp. It keeps me from becoming a prompt-engineer instead of an analyst.
One month without this, and I can feel the rust. My thinking gets sloppier because I'm not *doing* the work. I'm orchestrating AI to do it.
The Numbers: What This Actually Buys You
Let me be concrete about the time savings because that's what matters.
Before AI, my typical week looked like this:
- Data cleaning and prep: 6 hours
- Writing reports and summaries: 8 hours
- Email and admin: 3 hours
- Meetings and calls: 6 hours
- Actual thinking/analysis: 7 hours
After integrating AI smartly:
- Data cleaning and prep: 2.5 hours (AI + manual verification)
- Writing reports and summaries: 5 hours (drafting faster, thinking more)
- Email and admin: 1.5 hours (AI drafts, I review)
- Meetings and calls: 6 hours (unchanged — these need presence)
- Actual thinking/analysis: 12 hours (the gain)
That 5 extra hours of thinking time a week? That's where the real productivity is. Not in speed, but in depth. In having the mental space to actually understand what the data is telling me instead of just processing it.
That's not 10x faster. That's better faster. Subtle difference, but everything.
My Perspective
Back in my Economics M.A., one professor — I wish I remembered his name — said something that stuck: "The bottleneck is never the data. It's always the thinking."
At the time, I thought he meant analysis paralysis, overthinking. I was wrong. He meant cognitive capacity. Most of us aren't drowning in analysis. We're drowning in busy work that prevents analysis.
That's what I got about AI. It's not a replacement for thinking. It's a way to buy back the time thinking requires. And I was completely wrong about what would make me faster — it wasn't ChatGPT writing better emails. It was ChatGPT handling the annoying parts so I could spend 90 minutes thinking about why a fund's strategy might break in the next market crash.
What surprised me: the tools matter way less than the judgment about which tools to use when. I've met people using every cutting-edge AI available and producing worse work than they did before. And I've met people using ChatGPT for 20 minutes a day and shipping significantly better output.
The difference wasn't the tool. It was whether they used it to avoid thinking or to enable thinking.
Final Thoughts
If you're waiting for AI to make you 10x more productive, stop. It won't. What it might do is make you 30% faster at dumb tasks so you have energy for the smart ones.
And honestly? That's enough.
The people who are actually winning with AI aren't the ones using it the most. They're the ones using it *correctly* — which means knowing exactly what it's good for (automation, drafting, research) and what it's not (judgment, original thinking, client trust).
If you're in a job like mine — data, analysis, writing — start small. Pick one repetitive task. Use AI for 2-3 weeks. Measure the time saved. Measure the quality. If it's real, scale it. If it's not, move on.
And remember: the goal isn't to be faster at everything. It's to be faster at the stuff that doesn't matter so you can be better at the stuff that does.
That 90-minute train ride back to Kalyan? I'm still reading about markets and thinking about ideas. AI hasn't changed that. But it's given me an extra 5 hours a week to have thoughts worth thinking.
That feels like the actual win.
Dattatray Dagale
Data Analyst • Blogger • Mumbai
I'm a data analyst from Kalyan, Maharashtra, working at Morningstar. I write about personal finance, career growth, and everyday life for Indian millennials — the stuff I wish someone had told me earlier.
Written by Dattatray Dagale • 01 October 2026
.png)
.png)
0 Comments