SevyDevy

Free courseGenAI · Beginner · No maths · No coding needed

GenAI from ScratchThe free Generative AI course for beginners

8 modules · 32 lessons · 5½ hours · Free to read · Written lesson by lesson · By Saket Bhatnagar

Researched and drafted with AI, reviewed and edited by Saket Bhatnagar.

In short

GenAI from Scratch is a free Generative AI course for beginners, in 32 lessons across 8 modules. It explains how AI chat apps really work, from tokens and next-word prediction to prompting, RAG, agents, images and voice, and how real AI products are built, costed, tested and kept safe. Every lesson ends with a question to check yourself.

Lessons
32
Modules
8
Hands-on demos
33
Checks
82
Start the course, freeSee the syllabus

Lala video lessons

Every lesson will also come as a video taught by Lala, your AI mentor, personalized to you. The written lessons are here to read today.

Who it is for

  • Students and freshers who keep hearing "GenAI" in placement talks.
  • Working engineers who use AI tools but want to know what happens inside.
  • Product managers, designers and founders who need to make AI decisions.
  • Anyone curious. No maths and no coding needed until the optional bits of module 7.

What you will be able to do

By the last lesson.

  1. Explain how an AI chat app turns your question into an answer, in words a friend would follow.
  2. Write prompts that get clear, useful and correctly formatted answers the first time.
  3. Know why AI makes things up, and the ways to make it more truthful.
  4. Understand RAG, embeddings and vector search, and when to use them instead of fine-tuning.
  5. Describe how AI agents use tools, and where they need a human to say yes.
  6. Sketch a real AI product: its parts, its monthly cost, how to test it and how to keep it safe.

How every lesson works

Built so the idea sticks, not just so you finish.

  1. A story firstEvery idea starts with something from everyday life: a chai stall, an exam, a cricket match.
  2. Then the real ideaThe technical word, explained in plain English, with a diagram that shows how it works.
  3. Try it yourselfPlay with the idea right in your browser, with a small demo built for the lesson.
  4. Check yourselfA short question at the end. Every answer, right or wrong, tells you why.
  5. Watch with LalaA short animated video for every lesson, narrated by Lala. Coming lesson by lesson.

Try one now

This is a real demo from lesson 2.1. Type anything and watch how an AI reads it.

Try it yourselfSee your words the way AI sees themAI reads text as small pieces called tokens. Try the Hindi and emoji examples too.

Modules

8 modules and 32 lessons. Read in order, or open any lesson.

  1. Module 1 · 4 lessons · 39 min

    What GenAI really is

    What makes AI "generative", where it sits in the AI family, and how a machine learns from examples.

    1. What is generative AI?
    2. AI vs machine learning vs deep learning vs GenAI
    3. How a machine learns from examples
    4. What generative AI can and cannot do

  2. Module 2 · 8 lessons · 81 min

    How a language model thinks

    Tokens, next-word prediction, embeddings, attention, training, temperature, context windows and hallucinations.

    1. Tokens: how AI reads your words
    2. Next-word prediction: the one trick behind chat AI
    3. Embeddings: meaning as a map
    4. Attention and transformers, without maths
    5. How language models are trained
    6. Temperature: the creativity dial
    7. Context window: how much AI can keep in mind
    8. Why AI makes things up (hallucinations)

  3. Module 3 · 5 lessons · 48 min

    Talking to AI: prompting

    How to brief AI like a great manager: context, examples, step-by-step thinking and clean formats.

    1. A prompt is a brief to a new intern
    2. Show, don’t tell: examples in prompts
    3. Let it think step by step
    4. Getting clean, structured answers
    5. Prompt mistakes and how to fix them

  4. Module 4 · 4 lessons · 41 min

    Making AI know your stuff

    Why AI does not know your data, and how RAG, chunking and vector search fix that.

    1. Why AI doesn’t know your data
    2. RAG: the open-book exam
    3. Chunking and vector search
    4. Prompting, RAG or fine-tuning?

  5. Module 5 · 3 lessons · 31 min

    AI that takes action: agents

    Tools, function calling, the agent loop and MCP.

    1. Tools: giving AI a phone and a calculator
    2. Agents: think, act, check, repeat
    3. MCP: one plug for every tool

  6. Module 6 · 2 lessons · 20 min

    Beyond text: images, voice and video

    How AI makes pictures from noise, and how voice and multimodal assistants work.

    1. How AI makes images
    2. Voice and multimodal AI

  7. Module 7 · 3 lessons · 32 min

    Building real AI products

    The parts of an AI app, what it costs, and how to know it is good.

    1. The anatomy of an AI app
    2. Cost and speed: the token meter
    3. How to know your AI is good

  8. Module 8 · 3 lessons · 38 min

    Safe, fair, and your career

    Safety, bias, privacy and prompt injection, then careers, interview questions and your capstone.

    1. Safety, bias, privacy and prompt injection
    2. GenAI careers and interview questions
    3. Capstone: design your own AI helper

Why this course

An honest look at the usual ways to learn it.

How GenAI from Scratch compares with the usual ways to learn GenAI
Way to learnPricePlain languageHands-on in the browserCheck after every lessonBuilds up to real AI apps
GenAI from ScratchFreeYes, everyday examples firstYes, a demo in most lessonsYes, with a reason for every answerYes: RAG, agents, cost, tests, safety
Big-brand intro coursesFree to watch, certificate often paidMostlyRarelyShort quiz at the endUsually stop at the basics
Paid bootcamps₹20,000 to ₹2,00,000+Often jargon-heavyProjects, outside the lessonAssignmentsYes
Video marketplace courses₹500 to ₹4,000Varies by teacherCode-along videosSometimesOften tool-first, dated fast
YouTube playlistsFreeVariesNoNoScattered, no clear order

Typical prices and features as of October 2026; individual courses vary. A free certificate is planned for this course.

Questions people ask about this course

Questions people ask

Is this generative AI course really free?
Yes. Every lesson, diagram, demo and check is free, with no trial, no card and no account needed. Open any lesson and start.
Do I need to know coding or maths?
No. Modules 1 to 6 use everyday examples and in-browser demos, not formulas. Module 7 shows a little optional code for people who want to build.
How long does the course take?
About five and a half hours in total. Each lesson takes 8 to 15 minutes, so one lesson a day finishes the course in about a month.
Will there be videos?
Yes. Lala, SevyDevy’s AI mentor, will narrate an animated video for every lesson. The written lessons and demos are ready first, and videos are added lesson by lesson.
Is there a certificate?
Not yet. A free certificate with a final check is planned for this course.
What should I do after the course?
Practise. Try a GenAI mock interview with Lala, design an AI system on the system design board, and build the capstone helper from the last lesson.

Lesson 1.1 · 10 minutes

Start with “What is generative AI?”

Free, no account, no card. Open it and start reading.

Start lesson 1