Timeless Wisdom. Better Products.

The AI PM Gita

Eighteen lessons on building AI products, each drawn from the world's oldest epic.

No PhD Required · A Field Manual, Not a Theory Book

Three thousand years ago, the Mahabharata asked every question that haunts an AI product team today: When do you trust the powerful thing you don't fully understand? What does a confident wrong answer cost? Who is accountable when the machine speaks in your name? This book takes eighteen of the epic's sharpest moments and turns each into a working discipline for anyone building products in the age of AI.

Get the Book See the 18 Parvas
"Truth is not what the model produces. Plausibility is." — Parva 2 · The Clay Bird
The Premise

An ancient epic. A brand-new job.

The Mahabharata figured out product management three thousand years before we gave it a job title.

A wise counselor who sees everything and is ignored — that is every PM who was never given the authority to say no. An archer who sees only the eye of the bird — that is precisely how a language model sees the world, brilliant inside its training and blind outside it. A queen who asks one direct question while a hall full of experts answers everything except what she asked — that is every roadmap meeting where the user wanted their Tuesday back and got a chatbot instead.

Each of the eighteen Parvas opens with one of these stories, told properly — then refuses to leave it as a metaphor. The story becomes a framework, the framework becomes field exercises, and the exercises become a seven-day action plan you can start Monday. Story first. Discipline always.

You need exactly twelve words of AI vocabulary to read this book, and Parva 2 hands them to you in one table. What the book actually teaches is the part no vocabulary can carry: judgment — what to build, what to refuse, what a mistake costs, and who protects the human it lands on.

Written for

  • New & aspiring PMs stepping into AI without an engineering past
  • Working PMs whose products suddenly grew a model inside them
  • Engineers & designers moving toward product ownership
  • Founders & builders deciding where AI belongs — and where it doesn't
  • Anyone who loves a great story and distrusts hype

The Shift Right — a manual for now, and for what's coming

Status chasing · ticket writing · coordination theater Customers · discovery · judgment · evaluating AI itself

One thesis runs through all eighteen chapters: as AI agents absorb the coordination work and PM and engineering roles converge, the center of gravity of product management moves toward the customer. The left side of the job is leaving — with or without your permission. Every framework in this book trains a right-side muscle, because that side is becoming the entire job.

Inside every Parva

॥1॥

The Story

A moment from the epic, told with the care it deserves — the blind king, the clay bird, the spiral formation.

॥2॥

The Lesson

The product discipline hiding inside it, in plain language an average techie can enjoy.

॥3॥

The Framework

A named, reusable tool — the Drona Audit, the Three Rings, the Build Ladder — ready for Monday's meeting.

॥4॥

The Practice

Field exercises, a 7-day action plan, key takeaways, and a leader's story held to a strict evidence standard.

The Skeleton

Eighteen Parvas. Six movements.

From "what is a model" to "what is your oath" — built to be walked in order, or entered wherever the fire is.

I

The Awakening

Parvas 1–4
01See What Leadership CannotVidura & the blind king
02What AI Actually Is (No PhD Required)Drona's clay bird
03The Map of AI: LLMs, Agents, Models, Oh MyThe eleven armies
04The 3 Sins of an AI PMDuryodhana's gates
II

The Discovery

Parvas 5–7
05Jobs to Be Done, Not Features to Be ShippedDraupadi's question
06If a Spreadsheet Works, You Don't Need AIThe army or Krishna
07Your Data Is the Product Before the ProductScouting Kurukshetra
III

The Strategy

Parvas 8–10
08The AI Roadmap Is Not a Feature ListThe charioteer's counsel
09How Do You Know the AI Is Working?The eye of the fish
10You Probably Just Need the APIKarna's borrowed weapon
IV

The Execution

Parvas 11–13
11The PM's Secret Weapon: Prompt EngineeringSanjaya's divine sight
12Break Your Model Before the User DoesThe Yaksha's questions
13When the Model Fails AnywayThe Chakravyuha
V

The Leadership

Parvas 14–16
14How to Talk AI to Non-AI PeopleThe Vishwarupa
15The Uncomfortable Questions You Must Ask FirstThe wheels touch earth
16Earning Your Seat at the ML TableThe twin princes
VI

The Mastery

Parvas 17–18
17The AI PM Stack: What You ControlBhishma's oath
18Your One Job: Keep the Human at the CenterThe dog at heaven's gate
"Nobody wants a chatbot. They want their Tuesday back."Parva 5
"Your model's words are your company's words."Parva 4
"The first is rent. The second is equity. Only one compounds."Parva 17

Srinivas Mullapudi

Srinivas has spent two decades building data platforms and AI systems inside large enterprises — the rooms where the hype meets the invoice. He is the author of the Apicbook Ramayana series for children, and writes about AI, product, and the epics on his Substack. This book is what happened when the two halves of his bookshelf finally met.

Begin at Parva 1 — or wherever the fire is.

The book includes a crisis index: whatever is burning this week — a demo that fooled the room, a launch with no eval, a stakeholder who wants magic — there is a Parva for it.

Get the Book Read the Substack