The Productivity Problem
Most knowledge workers spend a disproportionate amount of time on low-value, high-volume tasks: drafting routine emails, reformatting documents, summarising meetings, compiling reports from scattered notes. These tasks are not intellectually demanding — they are just time-consuming. ChatGPT can handle the mechanical part of almost all of them, freeing you for the thinking that actually requires a human.
This chapter builds a complete personal productivity system using ChatGPT. You will see exactly what to type, what to expect back, and how to connect the pieces into a daily workflow.
The Core Principle: ChatGPT as a Writing and Thinking Partner
ChatGPT is most productive when you treat it as a tireless assistant who:
- Knows your context (because you told it)
- Does a first draft immediately
- Accepts revisions without ego
- Works at any hour
The shift in mindset is this: stop writing from scratch, start editing from a draft. Even a mediocre first draft from ChatGPT is faster to edit than staring at a blank page.
Setting Up Custom Instructions
Before building any workflow, configure Custom Instructions so ChatGPT knows your context without you having to re-explain it every session.
How to enable:
- Click your profile icon → Settings → Personalization → Custom Instructions
- Fill in two fields:
Field 1: What ChatGPT should know about you
I am a mid-level marketing manager at an e-commerce startup in Bengaluru.
We sell home decor products on Flipkart, Amazon India, and our own website.
My team has 6 members. I report to the VP of Marketing.
I work in Indian English (not American English).
Key tools I use: Google Workspace, Slack, Notion, Razorpay dashboards.
Field 2: How ChatGPT should respond
Be concise. Prefer bullet points over long paragraphs.
Use professional but approachable language — not overly formal.
Default to Indian business context (₹ for currency, Indian date format DD/MM/YYYY).
When I ask for an email, give me the email directly — no preamble like "Sure! Here is an email...".
Always suggest one or two follow-up improvements at the end of your response.
With these in place, every subsequent conversation inherits your context. You no longer need to say "I work in marketing in Bengaluru" every time.
Building a Daily Workflow
Here is a complete day-in-the-life workflow using ChatGPT as your productivity partner.
Morning Brief (5 minutes)
Each morning, paste a short status update into ChatGPT and ask it to structure your day.
Prompt:
Today is Wednesday 02/07/2026.
Here are my raw priorities:
- Quarterly review deck due Friday
- Follow up with Anjali about the Flipkart ad spend report
- 3 PM call with the content agency about Diwali campaign
- Respond to 12 pending emails (most are routine approvals)
- My team member Rahul asked for leave on Thursday
Draft a structured daily plan with time blocks. I have 8 AM to 6 PM available.
Mark high-priority tasks with [P1] and keep meetings fixed.
What you get back:
08:00 – 09:00 [P1] Deep work: Quarterly review deck (slides 1-10)
09:00 – 09:30 Email triage — batch-process the 12 pending approvals
09:30 – 10:00 Follow up with Anjali re: Flipkart ad spend report
10:00 – 12:00 [P1] Deep work: Quarterly review deck (slides 11-20, add data)
12:00 – 13:00 Lunch + break
13:00 – 14:30 Content agency prep: review Diwali brief, prepare 3 questions
14:30 – 15:00 Buffer / catch-up emails
15:00 – 16:00 [Fixed] Call with content agency
16:00 – 17:00 Address Rahul's leave request, update project coverage plan
17:00 – 17:30 End-of-day wrap-up: update Notion task board
This takes 30 seconds to generate and gives you a structured starting point. Edit where needed.
Email Triage
For bulk email processing, paste the subject lines and a one-line summary of each email and ask ChatGPT to categorise and draft responses.
Prompt:
Here are 5 emails I need to respond to today. For each, tell me:
- Category (Approval / Information needed / FYI / Action required)
- Suggested response (2-3 lines maximum)
Emails:
1. Subject: Leave request — Rahul Singh (Thu-Fri)
Summary: Rahul wants 2 days leave, his current sprint tasks are 70% complete.
2. Subject: Invoice #4521 from XYZ Agency — ₹85,000
Summary: Agency sent their June invoice, needs approval to send to accounts.
3. Subject: Flipkart Big Billion Day — Ad slot booking deadline today
Summary: Flipkart account manager says we need to confirm ₹1.2 lakh ad budget by 5 PM.
4. Subject: Team lunch on Friday?
Summary: Colleague Priya is organising a team lunch, asking if Friday works.
5. Subject: Updated SEO report — June 2026
Summary: SEO agency sent monthly report, no immediate action needed.
ChatGPT returns a neat table with categories and draft responses. You copy-paste and adjust names or figures, saving 20-30 minutes of individual composition.
Meeting Prep
Before any important meeting, use ChatGPT to help you prepare.
Prompt:
I have a 1-hour call at 3 PM today with a content agency about our Diwali 2026 campaign.
Context:
- Budget: ₹8 lakh for all content (videos, static creatives, influencer briefs)
- Timeline: Content needed by 30 September for review
- Our brand voice: warm, aspirational, rooted in Indian culture — not generic "sale" messaging
- Last year's campaign underperformed on video content (low views)
Give me:
1. Three key questions I must ask the agency
2. Two potential red flags to watch for
3. A one-paragraph briefing note I can send to my VP before the call
In under 10 seconds, you have structured preparation that would normally take 30 minutes of thinking and writing.
End-of-Day Summary
At the end of each day, paste your messy task notes and ask ChatGPT to write a clean summary you can share with your manager or keep for records.
Prompt:
Here are my rough end-of-day notes. Convert them into a clean daily summary
in bullet-point format, grouped by: Completed / In Progress / Blocked / Tomorrow.
Notes:
finished most of the deck, slides 1-15 done, need data for slides 16-18
Anjali sent the report at 2 PM, reviewed it, found a discrepancy in week 3 numbers
agency call went well, they'll send revised quotes by Thursday
Rahul leave approved, assigned his tasks to Meena
still waiting for finance to approve the invoice
need to send Flipkart ad confirmation first thing tomorrow
Automating Repetitive Document Tasks
Standard Operating Procedures (SOPs)
Give ChatGPT a messy process description and ask it to convert it into a numbered SOP.
Prompt:
Here is how our team processes a return request from a customer.
Convert this into a clean SOP with numbered steps, responsible party in brackets,
and a notes column for exceptions.
Process: customer calls or emails, support agent checks order in our OMS,
if order is within 7 days and product is undamaged we approve,
send a return pickup request to the courier (Delhivery usually),
once item received in warehouse quality team checks it,
if ok refund is initiated to original payment method in 5-7 business days,
if damaged we escalate to senior agent who decides case by case.
Filling Report Templates
If you have a fixed report format (weekly marketing report, sales summary, etc.) and raw data, ChatGPT can fill the template for you.
Prompt:
Fill in the following weekly report template using the data I provide below.
TEMPLATE:
---
Weekly Marketing Report — Week of [DATE]
Total Spend: ₹[X]
Revenue Generated: ₹[X]
ROAS: [X]
Top Performing Channel: [CHANNEL]
Key Insight: [2-3 sentences]
Next Week Priority: [1-2 sentences]
---
RAW DATA:
- Google Ads: ₹45,000 spend, ₹1,80,000 revenue
- Meta Ads: ₹38,000 spend, ₹95,000 revenue
- Flipkart PLA: ₹27,000 spend, ₹1,10,000 revenue
- Total orders: 312
- Top product: Brass Diya Set (87 units)
- Week of 23-29 June 2026
Creating Reusable Templates
One of the highest-leverage uses of ChatGPT is generating templates once and using them forever. Ask ChatGPT to create a template, then save it in Notion or a Google Doc.
Email Templates Worth Creating
1. New vendor introduction email
2. Payment reminder (polite, then firm, then final)
3. Delay notification to a customer
4. Influencer collaboration proposal
5. Internal escalation email
6. Meeting request to a senior stakeholder
7. Rejection of an unsolicited proposal (polite)
8. Announcement of a policy change to the team
Prompt to create any template:
Create an email template for [SITUATION].
Use [PLACEHOLDER] markers for variable fields (e.g., [CUSTOMER NAME], [ORDER ID]).
Keep it under 150 words. Tone: professional but warm.
Checklist Templates
ChatGPT can generate detailed checklists for any recurring process.
Prompt:
Create a pre-launch checklist for publishing a product listing on Amazon India.
Include: content quality checks, image requirements, pricing checks,
inventory verification, keyword review, and compliance items.
Format as a checklist with sub-items where needed.
Worked Example: Automating a Weekly Report from Raw Notes
This is a complete, realistic example from start to finish.
The Situation
Rahul is a digital marketing executive at a Pune-based EdTech startup. Every Monday morning he spends 90 minutes compiling a weekly performance report from five sources:
- Google Ads dashboard (manual export)
- Meta Ads Manager (manual export)
- Google Analytics (traffic data)
- Razorpay dashboard (revenue)
- His own rough notes from the week
The report goes to his manager by 10 AM.
Step 1 — Collect Raw Data
Rahul copies the key numbers from each platform into a single note:
Week: 23-29 June 2026
Google Ads: spend ₹52,000, clicks 4,200, conversions 68, revenue ₹1,56,000
Meta Ads: spend ₹41,000, clicks 7,100, conversions 44, revenue ₹88,000
Organic (Google Analytics): sessions 18,400, new users 12,300, avg session 3m 20s
Razorpay: total collections ₹2,89,000 (includes ₹45,000 from EMI plans)
Top course sold: Data Science Bootcamp (39 enrolments)
Cancelled orders: 7 (mostly payment failures, not cancellations)
Notes: meta performance dipped mid-week when we paused the video ad
google ads ROAS improved after bid strategy switch to max conversions
organic traffic spike on Wednesday — probably because of the blog post on AI careers
Step 2 — Paste into ChatGPT with a Structured Prompt
You are a data-savvy marketing analyst writing a weekly performance report
for a senior manager. Use the raw data below to write:
1. A 4-row KPI summary table (Metric | This Week | vs. Last Week | Status)
— use "N/A" for "vs. Last Week" since I have not provided prior week data
2. A "Key Insights" section (3-4 bullet points, each with one actionable takeaway)
3. A "Risks & Watchpoints" section (2-3 items)
4. A "Next Week Plan" section (3 priorities)
Raw data:
[paste the raw data block above]
Step 3 — Review the Output
ChatGPT returns a polished, structured report in under 5 seconds. Rahul reads it, adds the "vs. Last Week" column manually from last week's report, and changes one insight that he knows from context is slightly off (the organic spike was actually from a backlink, not the blog post).
Total time: 8 minutes instead of 90 minutes.
Step 4 — Save as a Reusable Prompt
Rahul saves the prompt template in Notion under "AI Prompts" with a placeholder where the raw data goes. Every Monday, he fills in the numbers, pastes, and gets the report draft instantly.
Common Pitfalls
1. Being too vague in prompts "Write me an email" produces generic output. "Write a 100-word payment reminder email to a vendor who is 15 days overdue on a ₹75,000 invoice — polite but firm" produces something you can actually send.
2. Accepting the first draft without reading it ChatGPT sometimes gets specific numbers wrong if they are embedded in complex prose. Always scan numeric values, names, and dates before sending anything externally.
3. Not saving your best prompts The prompt that produces a great result today is gold. Save it. Use Notion, a plain text file, or even a dedicated Google Doc called "My ChatGPT Prompt Library".
4. Using it only for writing ChatGPT is also excellent for thinking — asking "what am I missing in this plan?", "what are the risks here?", or "give me three counterarguments to my proposal". These thinking uses are underutilised.
5. Over-automating sensitive communications High-stakes emails (to investors, legal notices, apology emails to important clients) should always be written and reviewed carefully by a human. Use ChatGPT for the draft, not the final send.
6. Ignoring the Custom Instructions feature Users who skip Custom Instructions re-explain their context in every conversation, wasting time and producing less tailored outputs. Setting it up once saves hours over time.
Practice Exercises
-
Configure Custom Instructions for yourself. Write Field 1 (who you are and what you do) and Field 2 (how ChatGPT should respond) based on your actual work context. Save screenshots of both fields.
-
Take any 5 emails sitting in your inbox right now. Write a triage prompt in the format shown in this chapter and run it. Measure how much time it saves compared to responding one by one.
-
Think of a recurring document you produce (weekly report, monthly summary, meeting notes, project update). Ask ChatGPT to create a fill-in-the-blank template for it. Save the template and the prompt together.
-
Use the morning brief prompt to plan your actual tomorrow. After the day is done, evaluate: how closely did you follow the plan, and what would you change in the prompt?
-
Recreate the weekly report automation example using your own made-up data for a business of your choice (a restaurant, a tutoring centre, a clothing brand). Produce the final report and note where you had to manually correct ChatGPT's output.
Summary
- Custom Instructions let you personalise every conversation with your professional context, role, and preferred response style — set them once and benefit permanently
- A structured daily workflow (morning brief, email triage, meeting prep, end-of-day summary) can save 1-2 hours each day on purely mechanical writing and organising tasks
- ChatGPT works best for first drafts — your job shifts from writing to editing, which is dramatically faster
- Reusable templates and prompt libraries are the highest-leverage investment: build them once, use them hundreds of times
- The weekly report automation pattern (collect raw data → structured prompt → review and correct → save prompt template) is transferable to almost any regular reporting task
- Never send ChatGPT output externally without reading it — numbers, names, and dates are the most common sources of error
- ChatGPT is not just a writing tool — use it as a thinking partner to stress-test plans, identify blind spots, and generate alternatives