How to Use AI at Work: Practical Examples for Every Role in an Indian Office
Short answer: To use AI at work, pick one repetitive writing, summarising or spreadsheet task, give a chat assistant such as ChatGPT, Claude, Gemini or Copilot clear context and a sample, then check every fact before it leaves your desk. Start with low-risk drafts, keep confidential data out of personal accounts, and follow your company's AI policy.
Most AI use in Indian offices right now is quiet and unplanned. Someone in accounts asks ChatGPT for a VLOOKUP, someone in HR rewrites a job description, a sales executive polishes a WhatsApp message before sending it to a client. That helps a little. The real gains come when you choose tasks deliberately, write better instructions, and build a habit of checking the output.
This guide walks through what AI is good and bad at, concrete examples for HR, accounts, sales, teaching and management roles, a table of rough time savings per task, a worked prompt with its output critiqued, a list of data you should never paste into an AI tool, and a 30-day plan to start. Each role section links to a deeper guide in this series.
Where does AI actually help in office work?
AI helps most with first drafts, summaries, rewriting for tone, turning messy notes into structure, and explaining formulas or concepts. It is weak at facts it cannot see, current rules and rates, numbers it has to calculate in its head, and judgement about your specific people and clients.
The tools most people use are chat assistants: you type an instruction (a prompt) and the model replies with text, a table or a formula. They can also make mistakes that sound confident, often called hallucinations: a made-up circular number, a wrong GST rate, a meeting decision nobody took. So the working rule is simple: AI drafts, you decide.
- Good fits: emails and WhatsApp replies, job descriptions, meeting summaries, report outlines, Excel formulas, first drafts of policies and SOPs, lesson plans, rewriting English for clarity, translating to and from Hindi.
- Use with care: anything with numbers, tax or legal references, people decisions such as hiring or appraisals, client-facing claims, and research where you need a source you can cite.
- Poor fits: final legal, tax or HR-compliance advice, decisions about individuals, anything you cannot check yourself, and anything involving data you are not allowed to share.
If you are still deciding which assistant suits which job, our breakdown of which AI tool to use for what covers the main options side by side.
In this guide: 21 related articles
- How to Draft a GST Notice Reply With AI, Safely
- Best AI Course in Delhi: No-Code, Coding or ML, and How to Choose
- ChatGPT for Real Estate Agents: Listings, Client Messages and Follow-ups for the Indian Market
- AI Skills for Non-Technical Professionals: A 4-Level Ladder with Proof for Employers
- Should I Tell My Boss I Use AI at Work? How to Raise It and Stay Safe
- ROI of AI Training: How to Measure It with Your Own Numbers
- AI for Small Business Owners: Using It for Quotes, Stock, Suppliers and Staff
- AI Resume Screening: How to Shortlist Candidates Fairly with ChatGPT
- AI for Sales: Follow-Ups, Call Notes and Objection Replies That Still Sound Like You
- Will AI Replace Office Jobs? What It Is Taking Over, Task by Task
- AI for Chartered Accountants: What a Small CA Firm Can Safely Hand to AI
- How to Train Employees on AI: Choosing a Format and Building a Plan
- AI for Teachers: Build Lesson Plans and Question Papers Faster, Without Losing Quality
- Company AI Usage Policy: A One-Page Starter Template for Indian Firms
- ChatGPT for Excel Formulas: Worked Examples on GST, Salary and Sales Sheets
- ChatGPT for Performance Reviews: From Your Notes to a Fair, Specific Appraisal
- ChatGPT for HR: Drafting JDs, Policies and Employee Letters Without the Usual Mistakes
- Corporate AI Training in Delhi NCR: What to Check Before You Book
- AI Training for School Teachers: How to Plan Sessions That Change Classroom Practice
- AI Meeting Notes: How to Choose a Note Taker and Get Reliable Action Items
- AI Course for Working Professionals: What to Learn First, Role by Role
AI in the workplace: examples by role
The fastest way to start is to look at what people in your role already do every week and find the parts that are mostly typing, rewriting or summarising. Here is how that looks in five common roles. The scenarios are illustrative, not real clients.
- HR and recruitment. Say you are the only HR person at a 120-person logistics company in Gurugram. AI can draft a JD from a hiring manager's rough notes, write interview question banks by competency, turn a policy into an FAQ for staff, and draft offer and rejection emails. Our guide to ChatGPT for HR professionals has prompts for each. Screening CVs with AI needs extra care around bias and consent, covered in how to use AI for resume screening responsibly, and appraisal season has its own guide on writing performance reviews with AI help.
- Accounts and CA firms. In a small accounts team, AI is most useful as a formula and drafting assistant: building reconciliation formulas, cleaning exported Tally data, drafting client reminders, and making first-pass audit checklists. See AI for chartered accountants for a task map, our walkthrough of getting Excel formulas from ChatGPT, and the cautious, step-by-step approach to drafting a GST notice reply with AI, where a CA must review every line.
- Sales and business development. A sales executive at a Noida packaging firm can use AI to research a prospect's website before a call, draft follow-ups after each meeting, summarise long email threads, and prepare objection-handling notes. Our guide on AI for sales follow-ups shows how to keep messages specific rather than generic, and brokers have a dedicated guide on ChatGPT for real estate agents covering listings and client messages.
- Teaching. A Class 8 science teacher can turn a chapter into a lesson plan, a worksheet at three difficulty levels and a marking rubric in one sitting, then fix what AI gets wrong. Start with AI lesson planning for teachers; school leaders planning staff sessions can read how to plan AI training for school teachers.
- Managers and team leads. A manager's time goes into meetings, status reports and reviews. AI can produce meeting summaries with owners and deadlines, turn five team updates into one weekly report, and draft the first version of a project plan. Our comparison of AI meeting notes tools for Meet, Zoom and Teams shows how to get reliable action items.
- Owners of small businesses. A shop owner in Lajpat Nagar or a clinic owner in Dwarka wears every hat. AI can draft product descriptions, reply templates, simple SOPs for staff and Google Business Profile posts. The guide to AI for small business owners lists where to start without spending much.
How much time can AI save per task?
For writing and summarising tasks, AI often cuts the time roughly in half, sometimes more, once you include the time spent checking. The table below gives rough estimates based on the kinds of tasks we see in our agency work since 2013 and in AI workshops. They are not measured study results. Your numbers will vary with the task, your prompt and how much checking the output needs.
| Task | Typical role | Without AI (rough) | With AI, including checking (rough) | Where the remaining time goes |
|---|---|---|---|---|
| Job description from manager's notes | HR | 45–60 min | 15–20 min | Confirming salary band, location, reporting line |
| Interview question bank for one role | HR | 60 min | 15–25 min | Removing generic questions, adding role-specific ones |
| Excel formula you don't know | Accounts | 20–40 min of searching | 5–10 min | Testing it on 10 rows before applying to all |
| Client document-request email | Accounts / CA | 15 min | 5 min | Checking the document list for that client |
| Post-meeting follow-up email | Sales | 15–20 min | 5–8 min | Correcting numbers, dates and what was promised |
| Weekly team status report | Managers | 60–90 min | 20–30 min | Checking figures, adding judgement on risks |
| Lesson plan plus worksheet | Teachers | 60–90 min | 25–35 min | Matching the syllabus, fixing errors, adjusting level |
| Summary of a 40-page report | Anyone | 60+ min | 15–20 min | Spot-checking claims against the original |
Two things shrink these savings in practice. First, a vague prompt produces a vague draft that takes longer to fix than writing from scratch. Second, people skip the checking step and then lose far more time correcting a mistake that went out. If you want to measure this properly for a team, the guide on measuring the ROI of AI training shows how to log before-and-after times.
Using ChatGPT at work: a prompt, its output and what to fix
A good work prompt has four parts: who you are, what the task is, the facts the AI must use, and the format you want. Here is a worked example. Say you are an operations executive at a Noida logistics company and a client's shipment to Pune is delayed by two days because of a truck breakdown.
- Weak prompt: "Write an email to a client about a delayed shipment."
- Typical output: a long, over-apologetic email addressed to 'Dear Valued Customer', which invents a reason ('unforeseen circumstances beyond our control'), promises a delivery date you never gave, and offers a discount you are not authorised to give.
- Better prompt: "I am an operations executive at a logistics company in Noida. Write a short email to Mr Mehta, purchase manager at a Pune distributor. Facts: consignment 4471 left Noida on Monday; the truck broke down near Vadodara; a replacement truck is arranged; new expected delivery is Thursday evening. Tone: direct and polite, Indian business English. Do not offer any discount or compensation. Under 120 words. End by offering a call with our dispatch head."
- What to check in the better output: the consignment number and day names match your system, it has not added a reason or promise you did not give, the name and designation are right, and the tone suits this client's history with you.
The pattern holds for almost every office task: give the facts, say what not to do, and set a length. Save prompts that work in a shared document so the next person in your team starts from a good version instead of a blank box.
Never paste this: what to keep out of AI tools
Treat a personal AI account like a public email address: assume anything you paste could be stored and seen outside your company. Unless your company has approved a specific tool and plan for this data, keep the following out:
- Identity numbers: Aadhaar, PAN, passport, voter ID, bank account and card numbers.
- Salary sheets, appraisal ratings, disciplinary records and medical information about employees.
- Client financials, GST data or ledgers that identify the client by name or GSTIN.
- Student records: names with marks, attendance, photos, health or learning-support details.
- Unreleased prices, tenders, bids, board papers and merger or funding discussions.
- Passwords, API keys, server details and source code covered by an NDA.
- Anything in a contract or NDA that says it must not be shared with third parties.
If you need AI help with such a document, replace real details with dummy ones (Client A, ₹X, GSTIN removed) and put the real values back yourself. As of November 2026, the business and enterprise plans of the major AI tools generally state that workspace data is not used to train their models by default, while consumer plans may use chats for training unless you switch it off. Check the current terms of the exact plan your company uses with your IT team, because these settings change. India's Digital Personal Data Protection Act, 2023 also places duties on organisations that handle personal data; for what it means for your company, ask your legal team or a lawyer rather than an AI tool.
For the rules a company should write down, see our guide to writing a company AI usage policy. If you are unsure whether to tell your manager you already use AI, read should you tell your boss you use AI first.
Which AI tools are worth using for office work?
For most office work, one general chat assistant plus whatever AI is built into your email and documents is enough. Here is how the common options fit. Features and plans change often, so treat this as a starting point as of November 2026 and check each tool's current pricing page.
| Tool | Where it fits at work | Watch out for |
|---|---|---|
| ChatGPT | General drafting, Excel help, brainstorming, rewriting | Personal vs business account settings; check facts it states |
| Claude | Long documents, careful rewriting, structured summaries | Same data rules apply; verify any figures |
| Gemini (in Google Workspace) | Gmail drafts, Docs, Sheets, Meet notes if your plan includes them | Features depend on your Workspace plan |
| Microsoft Copilot (in Microsoft 365) | Outlook, Word, Excel, Teams meeting recaps | Many features need a separate Copilot licence |
| Perplexity | Quick research with linked sources | Open the sources; summaries can still misread them |
| NotebookLM | Answers drawn only from documents you upload | Only as good as the documents you give it |
Pick the one your company already pays for or approves before buying anything new. As of November 2026, paid individual plans of the main assistants commonly cost somewhere around ₹1,700–2,000 a month, but this varies by plan and changes often, so check before you buy.
A 30-day plan to start using AI at work
You do not need a course to begin, but you do need a routine. This plan works for one person or a small team.
- Week 1: list your ten most repeated tasks and note roughly how long each takes. Mark the three that are mostly writing, summarising or spreadsheet work.
- Week 1: read your company's IT or AI policy, or ask HR whether one exists. Confirm which tools and accounts you are allowed to use.
- Week 2: try AI on those three tasks with dummy or non-sensitive data. Write one detailed prompt per task using the four-part structure above and save the versions that work.
- Week 2: for each output, note what you had to fix. That list tells you what to add to the prompt next time.
- Week 3: tell your manager what you are trying and show one before-and-after example. Agree on what is fine to use AI for in your team.
- Week 3: add one new task type, such as meeting summaries or a weekly report.
- Week 4: compare your time on these tasks with Week 1. Share your saved prompts with a colleague and keep only the ones that still work.
- Week 4: decide what to learn next: better prompting, a second tool, or a simple automation that moves data between apps.
People worried about where this leads can read our honest look at whether AI will replace office jobs, and the list of AI skills for non-technical professionals shows which abilities are worth building beyond basic prompting.
Rolling out AI to a whole team
Teams get more from AI when the company sets rules first, trains people on their own tasks, and measures the result. Individual experimenting tends to stall after a few weeks without that support.
Start with a short written policy, then run hands-on training where each person works on real tasks from their role, not a tour of twenty tools. The guide on how to train employees on AI covers formats, group sizes and follow-up, and companies comparing options locally can use our buyer's checklist for corporate AI training in Delhi NCR. If you want the details of what our workshops include, see the corporate AI training programmes page.
If you are choosing a course for yourself rather than a team, our guide to picking an AI course as a working professional compares the options. The Spettro School Applied AI Professional course covers these tools, prompting and no-code automation over 8 weeks of live evening or weekend classes.
Frequently asked questions
Is it okay to use ChatGPT at work?
Usually yes, if your company allows it and you keep confidential data out. Check your IT or AI policy, use an approved account if one exists, and review everything before it goes to a client or manager.
Can my company see what I type into ChatGPT?
On a company-managed business account, administrators may have access to usage and data depending on the plan. On company laptops and networks, IT can often monitor activity. Assume work devices are not private.
What is the best AI tool for office work?
The one your company already approves and pays for is usually the right start, often Copilot in Microsoft 365 offices or Gemini in Google Workspace offices. For general drafting, ChatGPT and Claude are both capable; pick one and learn it well.
Do I need coding skills to use AI at work?
No. Chat assistants work in plain English or Hindi. Coding helps only if you want to build custom tools; for automations between apps, no-code tools like Zapier and Make are enough for most office needs.
How long does it take to get good at using AI at work?
Most people get useful results within a week of daily use on real tasks. Getting consistently good, with reusable prompts and a habit of checking, typically takes a month or two of practice.
Keep reading in AI at work
- AI Meeting Notes: How to Choose a Note Taker and Get Reliable Action Items
- AI Course for Working Professionals: What to Learn First, Role by Role
- AI Training for School Teachers: How to Plan Sessions That Change Classroom Practice
- Corporate AI Training in Delhi NCR: What to Check Before You Book
- ChatGPT for HR: Drafting JDs, Policies and Employee Letters Without the Usual Mistakes
Courses: Applied AI Professional · Corporate AI training · All guides in AI at work