How to Learn AI in India: A 90-Day Roadmap (2026)
The fastest way to learn AI in India is not to start with maths or a two-year degree — it is to start using AI tools on your real work this week, then go deeper in a structured way over about 90 days. This roadmap splits those 90 days into three phases and gives you two tracks: a non-tech track for working professionals, students and business owners, and a tech track for people who can (or want to) write code.
Why a 90-day plan works
Most people who "try to learn AI" stall for one of two reasons: they jump into theory that feels disconnected from their job, or they watch endless videos without building anything. A 90-day plan fixes both problems:
- It is long enough to move from curiosity to a real, demonstrable skill.
- It is short enough to keep momentum alongside a job or college.
- It is project-driven, so at the end you have something to show an employer or client, not just a certificate.
Plan for roughly 45–60 minutes a day on weekdays, plus one longer session on the weekend. If you can only manage less, stretch each phase — the order matters more than the speed.
Before you start: pick your track
| Non-tech track | Tech track | |
|---|---|---|
| Who it is for | Marketing, finance, HR, operations, sales, teachers, founders, students from any stream | Developers, engineering students, data analysts comfortable with code |
| Main goal | Use AI tools to do your current work faster and better | Build AI-powered applications and automations |
| Coding needed | None | Basic Python (you can learn it during the plan) |
| End-of-plan project | A documented AI workflow that saves you hours every week | A working LLM app (for example, a chatbot over your own documents) |
Both tracks share Phase 1. They split in Phase 2.
Phase 1 (Days 1–30): AI foundations for everyone
The goal of the first month is fluency: you should be comfortable using at least two AI assistants and understand, at a practical level, what they are good and bad at.
Week 1: Understand what AI actually is
- Learn the basic vocabulary: AI, machine learning, large language models (LLMs), generative AI, prompts, tokens, hallucinations.
- Understand the key limitation early: AI assistants can produce confident, fluent answers that are wrong. Verification is part of the skill.
- Create free accounts on two assistants (for example ChatGPT and Gemini) and use them for small, everyday tasks.
A good shortcut for this week is the free AI in 60 Minutes course, which covers these basics in a single sitting so you can spend the rest of the week practising.
Week 2: Use AI on your real work
Pick three tasks you do repeatedly and try AI on each one. Examples:
- Drafting and replying to emails (in English or Hindi, or translating between them)
- Summarising a long PDF, policy document or meeting transcript
- Creating a first draft of a presentation outline or report
- Explaining a concept you are studying in simpler language
Keep a simple log: task, tool used, prompt, what worked, what you had to fix.
Week 3: Learn to write better prompts
This is the single highest-return skill in the whole roadmap. Practise:
- Giving the AI a role and context ("You are a finance analyst at a mid-sized Indian manufacturing firm…")
- Being specific about format (table, bullet points, word limit)
- Providing examples of what good output looks like
- Iterating — treating the first answer as a draft, not the final result
Week 4: Explore the wider toolkit
Try AI inside the tools you already use — Microsoft Copilot in Office apps, Gemini in Google Workspace, AI features in Canva or Notion. Note which ones genuinely save time for you.
If you want a structured version of Weeks 2–4, AI Essentials: ChatGPT, Gemini & Copilot for Everyday Work walks through these tools with workplace examples.
Phase 1 checkpoint: you can name three tasks where AI saves you real time, and you know how to check its output.
Phase 2 (Days 31–60): Go deeper on your track
Non-tech track: become the "AI person" in your role
Choose one area and go deep rather than wide:
- Data and reporting: use AI with Excel or Google Sheets to write formulas, clean data, build pivot tables and explain trends. Move on to AI-assisted dashboards in Power BI.
- Content and marketing: build a repeatable process for blog posts, LinkedIn content, ad copy and visuals, with your own brand voice.
- Automation: connect AI to everyday apps (email, sheets, forms, WhatsApp Business, CRMs) using no-code tools, so routine work runs on its own.
- Freelancing or a side business: package an AI-assisted service — content writing, data cleanup, social media management — and find your first clients.
Weekly rhythm for this phase:
- Learn one new technique (2–3 sessions).
- Apply it to a real task at work or for a real person (2 sessions).
- Document the before/after — time taken, quality, what you changed (weekend).
Tech track: from user to builder
- Weeks 5–6: Python and APIs. If you are new to Python, focus on the essentials: variables, functions, lists and dictionaries, reading files, and calling a web API. Then make your first call to an LLM API and print the response.
- Week 7: Prompting in code. Learn system prompts, structured (JSON) outputs, and how to handle errors, rate limits and costs.
- Week 8: Retrieval-augmented generation (RAG). Learn how embeddings and vector search let an LLM answer questions about your own documents — the most common pattern in real company projects.
Use free tiers and small test budgets while learning, and never commit API keys to public code repositories.
Phase 2 checkpoint: non-tech learners have one workflow that reliably saves time; tech learners have a script that answers questions from a set of documents.
Phase 3 (Days 61–90): Build, ship and show
The last month is about turning skill into proof.
Non-tech track capstone
Pick one project and finish it properly:
- An AI-assisted monthly reporting process for your team, with a written step-by-step guide
- A content system that produces a month of posts for a real brand or your own profile
- A no-code automation that, for example, reads incoming enquiry emails, summarises them and logs them to a sheet
Tech track capstone
Build and deploy a small but complete application:
- A chatbot over a college's or company's public documents
- A tool that extracts structured data from invoices or resumes
- A simple agent that uses tools — search, a calculator, a database — to complete a multi-step task
Deploy it somewhere others can try it, and write a short README explaining the problem, approach and limitations. Generative AI Mastery: Build LLM Apps, RAG & Agents with Python covers this full path, from API basics to RAG and agents.
Both tracks: make it visible
- Write a short LinkedIn post or case study about your project: the problem, what you built, the result.
- Add the project to your CV under a "Projects" or "AI skills" section with concrete outcomes.
- Share your work with your manager or in relevant communities — visibility is often what turns a skill into an opportunity.
A quick 90-day summary
| Days | Non-tech track | Tech track |
|---|---|---|
| 1–30 | AI basics, everyday tools, prompting | Same |
| 31–60 | Deep dive: data, content, automation or freelancing | Python, LLM APIs, RAG |
| 61–90 | Capstone workflow + case study | Deployed LLM app + README |
Common mistakes Indian learners make (and how to avoid them)
- Collecting certificates instead of projects. Recruiters and clients respond to things you have built and results you can explain.
- Starting with heavy theory. Linear algebra and neural network maths matter if you want to become an ML researcher. For most jobs that use AI, practical application comes first.
- Using only one tool. Different assistants have different strengths; knowing two or three makes you more effective.
- Pasting sensitive data into AI tools. Learn your employer's policy and avoid sharing confidential client or personal data with public tools.
- Learning alone with no deadline. Join a cohort, a study group, or set a public goal to keep yourself accountable.
Frequently asked questions
Can I learn AI in India without a technical background?
Yes. Most of the value AI creates at work today comes from people using AI tools well in their existing roles — in marketing, finance, HR, operations, teaching and business. The non-tech track in this roadmap needs no coding at all, only consistent practice on real tasks.
Do I need to know maths or coding to learn AI?
Not to use AI effectively. If you want to build AI applications, basic Python is enough to start with modern LLM APIs. Deeper maths becomes important only if you aim for research or core machine learning engineering roles.
Is 90 days really enough to learn AI?
Ninety days is enough to become genuinely useful with AI and to have a project that proves it. It is not enough to master the whole field — nobody does that — but it gives you a strong base to keep building on as tools change.
What is the best first step if I am a complete beginner?
Start with a short, free introduction so you understand the basic ideas, then immediately use an AI assistant on one real task from your work or studies. The free AI in 60 Minutes course is designed for exactly this first step.