It was 9:47 PM on a Tuesday. I was sitting in the Kalyan station, waiting for the 10:15 local to Dadar, laptop open on my knees, staring at a spreadsheet that needed cleaning. Forty-five minutes of data entry work. Forty-five minutes I didn't have.
I was supposed to be home by 11 PM. I had a personal finance article due the next morning. My mom had called twice asking when I'd eat dinner. And somewhere in that chaos, I was also supposed to review quarterly fund performance reports for work.
That night, instead of manually cleaning the data, I spent seven minutes writing a prompt for Claude. Seven minutes. The AI handled four hours of work.
I'm not saying this to brag. I'm saying it because most of us are still using AI wrong — or not using it at all. We treat it like a novelty. A ChatGPT tab we open when we're bored. We don't treat it like the actual time-multiplication machine it is. And that's costing us real hours, real money, real peace of mind.
Here's what I've learned in the last 18 months about actually being 10x more productive with AI. Not the hype version. The real version.
The Problem Isn't That You're Not Skilled — It's That You're Still Manual
Before I start, let me be honest about something: I used to think productivity meant discipline. Wake up at 5 AM. Have a routine. Check off boxes. Follow GTD. I read all the books. I tried Notion, then Todoist, then Things 3.
Spoiler: I was still slow.
The issue wasn't my discipline. The issue was that I was spending 80% of my time on work that a machine could do in 20% of the time. And I was proud of it. "Look how hardworking I am," my brain would say, while watching me manually format data at 10 PM on a local train.
Here's the thing about productivity culture in India — we've been conditioned to believe that more hours = more output. More struggle = more success. But that's just not true anymore. Not since AI.
The Real Productivity Isn't About Doing More
It's about doing what only you can do.
I can't delegate critical thinking to AI. I can't automate insights about whether a fund is worth recommending to our users. I can't outsource the decision about whether to change how we present risk metrics. Those require my brain, my experience, my judgment.
But I can absolutely automate the grunt work that leads to those insights.
At Morningstar, I spend maybe 15% of my week on analysis that actually matters — understanding patterns, making recommendations, challenging assumptions. The other 85% used to be data wrangling. Cleaning CSVs. Formatting reports. Chasing down missing data points. Manual quality checks.
That 85% is now maybe 20%. Everything else? AI handles it.
What Changed for Me
I stopped thinking of AI as a helper. I started thinking of it as a skill translator.
Here's what I mean: I'm good at spotting patterns in data. That's my skill. But I'm slow at writing the code to extract that data. I used to spend two hours writing Python to pull data from a messy source. Now I spend ten minutes describing what I need to Claude, and it writes the code. Then I spend the other 110 minutes actually thinking about what the data means.
That's a 10x difference. But not because I got 10x faster. Because I started using AI to bridge the gap between what I'm thinking and what I need to execute.
The Three Types of Work — and How to Automate Each
Not all work is created equal. And the way you use AI depends on what kind of work you're doing.
Type 1: Repetitive, Low-Brain Work
This is your data entry. Your email sorting. Your formatting. Your transcribing. Your summarizing.
This is also the easiest to automate. And the one where you get immediate time back.
How to use AI: Use it directly on the task. No framing needed. If you need to convert 200 LinkedIn profiles into a CSV with Name, Company, Role, Email — prompt an AI to write a script, or use Make (formerly Integromat) with Claude integration to do it automatically. If you need to summarize meeting notes daily, use Fireflies.ai. If you need to sort your emails by priority, use Gmail filters (not AI, but same effect — it's automation, which is what matters).
Real example from my life: I used to spend 30 minutes every Friday organizing my expense receipts from the week into categories for tax filing. Now I take a screenshot of my phone's CRED and PhonePe transactions, paste them into Claude with a template, and it organizes them into Investment, Tax-Deductible Work, Personal, and Other. Takes three minutes.
That's 27 minutes back per week. That's 1,400 minutes a year. That's 23 hours. That's two full days of work I'm not doing.
Type 2: Creative Work That Needs Structure
This is writing. Content creation. Design concepts. Strategy sketches.
This is where people get confused. They think AI can replace their thinking. It can't. But it can replace the friction between thinking and creating.
I write a weekly newsletter for a personal finance audience (yes, in addition to my day job — I know, I'm slightly mad). The bottleneck used to be the first draft. Staring at a blank page. Trying to find the right structure. Spending 90 minutes to write 800 words.
Now here's my process:
I spend 20 minutes thinking about the idea and writing three bullet points of what I want to say. I paste that into Claude with my writing style (I've built a custom instruction set in Claude that knows I'm Dattatray, that I use Indian context, that I write in first person, that I hate generic advice). Claude writes a draft in my voice — 800 words, structured, with my idioms. Then I spend 40 minutes rewriting, fact-checking, adding specific numbers, personalizing it further.
Total: 60 minutes for something that used to take 90 minutes. But more importantly: I'm not spending time on the slog of first-drafting. I'm spending time on what only I can do — making it real, making it honest, making it mine.
How to use AI: Use it as your first-draft engine, not your final answer. Give it constraints (word count, tone, structure, audience). Use custom instructions to bake in your voice. Then iterate manually.
Type 3: Analysis and Decision-Making
This is the highest-leverage work. This is where AI helps you think better, not just move faster.
I spend maybe four hours a week on this at Morningstar. Looking at fund performance anomalies. Thinking through how to communicate risk to investors who are scared. Deciding whether our rating methodology needs updating. Challenging our own assumptions about what makes a good fund.
AI doesn't make these decisions. But it makes me faster at exploring the decision.
Example: We were trying to understand why certain equity funds had low volatility but high returns (which shouldn't happen in theory). I could have spent a day manually checking different hypotheses. Instead, I spent 90 minutes with Claude, asking it to help me systematically work through possibilities: survivorship bias? Selection bias? Data error? Luck? Different market regime?
We worked through each one. Tested each one. Narrowed down to the real cause (a combination of selection bias and luck, if you're curious). What would have taken me a full day took me 90 minutes, and I understood the problem deeper because I was forced to think through it with a tool that could challenge me.
How to use AI: Use it as a thinking partner. Ask it to play devil's advocate. Ask it to stress-test your assumptions. Ask it to generate alternatives you might not have thought of. The goal isn't to accept its answer — it's to think better because of it.
| Type of Work | Time Saved (Typical) | AI Tool | What Matters Most |
|---|---|---|---|
| Repetitive, Low-Brain | 60-80% | Make, Zapier + Claude, Fireflies | Automation (set it once) |
| Creative (Writing, Design) | 40-50% | Claude, ChatGPT, Perplexity | First-draft speed + custom instructions |
| Analysis (Thinking) | 30-40% | Claude, ChatGPT (deep thinking mode) | Quality of thought, not speed |
The Practical Setup That Actually Works
This is where most people fail. They have AI tools. They don't have a system for using them.
Step 1: Audit Your Week
Spend one week (just one!) tracking where your time actually goes. Not your calendar. Your actual time.
I did this and it was brutal. I spent 15 hours a week on things that could have been automated or delegated. Fifteen hours! That's almost two full work days.
Use a simple Google Sheet. Every task you do, note it down. Time spent. If a machine could theoretically do it. Rate it as "Repetitive", "Creative", or "Analysis".
After the week, you'll see patterns. My pattern was: 8 hours on data cleaning, 4 hours on email management, 2 hours on formatting reports, 1 hour on meeting notes.
Step 2: Prioritize by Time × Frequency
Don't automate everything. Start with high-frequency tasks that take meaningful time.
For me: Data cleaning was 8 hours once a month = low frequency, high time. Not the best target. But email management was 4 hours a week = high frequency, meaningful time. That's a target.
I set up three Gmail filters with Claude integration to automatically label and prioritize emails. Took 30 minutes to set up. Now saves me 3-4 hours a week.
Step 3: Build Custom Instructions for Your Voice
This is critical if you're using AI for creative or communication work.
In Claude, I have a custom instruction that says:
"You are writing for Dattatray Dagale. He's a data analyst from Kalyan, Maharashtra, aged 32. He writes for Indian millennials aged 22-35 about money and careers. He uses first person. He's honest about uncertainty. He references Indian apps and context. He varies sentence length. He sounds like a real person, not an AI. He dislikes generic advice."
Now every time I use Claude for writing, it starts with my voice already embedded. I'm not rewriting everything. I'm just editing.
Step 4: Create Templates for Recurring Work
Every time you do a task more than twice, it's worth templating.
For my weekly newsletter, I have a template prompt:
"Write a 800-word personal finance essay for Indian millennials on [TOPIC]. Structure: hook story → problem → three solutions → real example → my opinion. Tone: honest, first-person, conversational."
I just fill in the topic. Claude does the rest. Then I edit.
For data cleaning at work, I have a template script that Claude wrote once, and now I just change the variables. Same for meeting summaries, expense categorization, and content ideas.
My Perspective
Here's what surprised me: I didn't become 10x more productive because I'm smarter or faster. I became 10x more productive because I stopped doing stupid work.
At Morningstar, we have dashboards that our fund managers look at weekly. I used to spend six hours manually updating these dashboards every Friday. Six hours! Of copy-pasting data from one sheet to another, formatting, checking for errors. Last year, I wrote a Python script with Claude's help. Now it runs every Friday morning automatically. Takes 30 seconds to verify.
I haven't gotten smarter. The work is the same. But now I'm not doing the stupid part. I'm spending those six hours actually thinking about what the data means, how to present it better, what questions our managers aren't asking but should be.
That's the real productivity gain. Not speed. Reallocation. Moving my time from "doing" to "thinking".
The other thing that surprised me: this isn't about having fancy tools. I use Claude, ChatGPT, Make, and Google Sheets with basic AI plugins. That's it. You don't need 47 different AI apps. You need three good ones, and you need to understand exactly what problem each solves. I see people with 20 AI tool subscriptions and they're not more productive — they're distracted.
And honestly? I was wrong about one thing. I used to think AI would make my job obsolete. "Why hire a data analyst if AI can do this?" I used to worry about that, especially when ChatGPT came out. Turns out, the opposite happened. AI made my job more valuable because now I can do work that actually matters. I'm not the person who cleans data. I'm the person who asks the right questions about what the data means.
Final Thoughts
That night at Kalyan station, when I automated four hours of work in seven minutes — that wasn't about me being smarter than the people sitting next to me on the local. It was about me being smarter about how I work.
You probably have 10-20 hours of automatable work in your week. Real hours. Not "optimize your morning routine" hours. Actual work that you could eliminate or reduce by 70% if you stopped pretending that manual work is the same as important work.
Start with one thing. One task that you hate. One task that's repetitive or that wastes time. Figure out how to automate it. Build the system. Then move to the next one.
You won't be 10x more productive because you have AI. You'll be 10x more productive because you stopped wasting time and started thinking about what actually matters.
That's the real productivity hack. And it's available to you right now.
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 • 06 July 2026
.png)
.gif)
0 Comments