.01
IISc Department of Management Studies

AI & The New
Management Playbook

Ashish Muralidharan

Product Manager, Microsoft

March 6, 2026

.02
The Opening

How many of you feel anxious about your future, your career, your relevance, because of AI?

Show of hands

.03

Act I

The Tension

.04

A one-way street going upwards

Volatility

The pace of change is accelerating. Markets, technologies, and business models shift overnight.

Uncertainty

The future is less predictable than ever. Historical data no longer guarantees reliable forecasts.

Complexity

Systems are deeply interconnected. A change in one domain cascades across the entire value chain.

Ambiguity

Cause and effect are unclear. Leaders must act with incomplete information and conflicting signals.

.05

Technology resets the cycle

01

Shipping & Navigation

Created global trade routes and built the British Empire. Whoever controlled the seas controlled the economy.

02

Industrial Revolution

Mechanized labor transformed agriculture and manufacturing. Entire populations migrated from farms to factories.

03

Information Technology

Digitized the world. Created the internet age and made knowledge workers the new elite labor force.

04

Artificial Intelligence

The next reset. For the first time, the technology can do the thinking, not just the lifting.

.06

Land, Labor, Capital, Technology

The classical factors of production have always been land, labor, and capital. Technology was the accelerant that made each more productive. But something has shifted. Technology is no longer just amplifying labor. It is learning to replace it.

Every previous wave automated the hands. AI automates the mind. When the tool can think, plan, and execute, the economics of labor change fundamentally.

We are at the outset of technology fundamentally replacing labor. Not augmenting it. Replacing it.

.07

Act II

The Pattern

.08

The same script, every cycle

01

Technology arrives

A new capability emerges that can perform tasks faster, cheaper, or at a scale humans cannot match.

02

Physical labor is displaced

The first wave hits manual work. Machines replace hands. Factories replace workshops. Robots replace assembly lines.

03

Operational labor is displaced

Process work goes next. Spreadsheets replace ledgers. Software replaces paperwork. ERP replaces middle management.

04

Intellectual labor is displaced

For the first time, AI targets the thinking itself. Strategy, creativity, analysis, judgment. This is where we are now.

.09

We may be running out of rungs

Physical
95%
Operational
72%
Intellectual
45%
???
15%

Every generation automated the rung below it and climbed higher. For the first time, AI is automating the rung we are standing on.

.10
The Ceiling

If you don't need intellectual labor anymore, what's left? You sell your time, or you sell your capital. And at that point, only capital owners have a role left in society.

The uncomfortable truth

.11

AI gets a body

The final frontier was always the physical world. AI could think, but it could not move. That barrier is falling.

Tesla Optimus — Humanoid robots being trained to perform warehouse and factory tasks autonomously.

Unitree — Agile quadruped and bipedal robots demonstrating parkour-level dexterity at consumer prices.

Hyundai Robotics — Industrial-scale robotic systems for construction, logistics, and heavy manufacturing.

Complete replacement of human labor across every dimension: intellectual, operational, and now physical.

.12

The Five Levels of AGI

Level Name Description Example
L1 Chatbots Conversational AI with no persistent memory across tasks ChatGPT (early), Siri, Alexa
L2 Reasoners Human-level problem solving in specific domains GPT-4, Claude, coding assistants
L3 Agents Systems that take actions on your behalf for hours or days Devin, Copilot Agents, Claude Code
L4 Innovators AI that generates novel ideas and scientific discoveries AlphaFold, AI-driven drug design
L5 Organizations AI that can do the work of an entire organization Hypothetical

We are at Level 3. Agents. And that has already changed everything.

.13

What is an agent?

A system that can pursue a goal by reasoning through a problem, planning a sequence of steps, using tools to interact with the real world, and continuously learning from feedback. It doesn't just respond. It acts.

Signal
Cognition
Action
Closure

The complete loop: perceive, think, act, and close. No human in the middle. That is the paradigm shift.

.14

Act III

Ground Zero: Software

.15

The perfect storm

100% Digital

Software lives entirely in the digital realm. No atoms to move, no physics to fight. AI can read, write, test, and deploy code without ever touching the physical world.

Loop Fully Automated

The signal-to-closure pipeline is entirely automatable. From reading a spec to shipping a feature, every step can be executed by an agent.

Rich Training Data

Billions of lines of open-source code, millions of Stack Overflow answers, decades of documentation. The training corpus for software is unmatched.

.16
The New Reality

Single engineers now spin up 20 agents that read a spec document and build the entire product end-to-end. No team. No sprint. No standup.

What's happening right now

.17

What the builders are saying

Andrej Karpathy

“There is a new kind of coding I call vibe coding, where you fully give in to the vibes, embrace exponentials, and forget that the code even exists.”

Naval Ravikant

“Product management is vibe coding. Engineering is prompt engineering. The lines between roles are dissolving.”

The vocabulary is changing because the work itself is changing. Titles matter less than capability.

.18
The Cost

The cost to build software
has come down to zero.

.19

Distribution is everything

When the cost of production drops to zero, the bottleneck shifts. Building is no longer the hard part. Reaching customers is. Every startup, every corporation, every solo builder now faces the same question: who will find your product?

This is classic Porter's Five Forces playing out in real-time. Barriers to entry collapse. Competition explodes. The only moat left is distribution, brand, and trust. Welcome to the Red Ocean.

The new job for everyone has become marketing. Production is commoditized. Distribution is the new scarcity.

.20

Already playing out

Block (Square): 50% layoffs — Jack Dorsey's fintech company cut half its workforce, citing AI-driven productivity gains as the reason fewer humans are needed.

IT services stocks down 25-30% — Indian IT bellwethers have shed a quarter of their market cap as investors price in a world where code is written by machines, not teams.

The Innovator's Dilemma in real-time — Incumbents are watching new entrants build in weeks what took them years. The playbook Christensen wrote about is happening at 10x speed.

This is not a future scenario. It is a present reality. The market is already repricing human labor.

.21

Act IV

The Economics

.22

What if things just get cheaper?

Remove Humans
Deflation
Goods Cheaper
Era of Abundance

If AI removes the human cost from production, the price of everything drops. Healthcare, education, legal services, software, content, logistics. The deflationary pressure could be unprecedented.

The optimistic case: abundance for everyone. The challenge: what happens to the people whose labor was the cost being removed?

.23

Where does value accrue?

Energy
Chips
Fabricators
Frontier Labs
Hyperscalers
App Layer
Distribution

The AI value chain mirrors the semiconductor value chain. Massive capital at the bottom (energy, chips), commoditizing layers in the middle, and disproportionate returns at the edges: whoever controls the infrastructure or the customer relationship wins.

Value concentrates at the extremes. Infrastructure (compute) and customer access (distribution) will capture most of the returns.

.24

Not in the model

Compute

Whoever owns the GPUs, the data centers, the energy contracts. This is the new oil, and it is fiercely concentrated among a handful of hyperscalers.

Context

Proprietary data, domain expertise, institutional knowledge. The model is generic. Your context makes it specific and valuable. This is the true moat.

Distribution

Access to users, trust, brand, existing workflows. When production is free, the ability to reach and retain customers becomes the only differentiator.

Agents don't need a UI. They need context, compute, and a channel to the customer. The model layer is commoditizing.

.25

Act V

From the Inside

.26

Top-down failed. Bottom-up worked.

The first attempt was mandates from leadership. Use AI. Integrate Copilot. Hit adoption metrics. It did not work. People used the tools to check a box, not to change how they worked.

What worked was giving every team unlimited compute and letting them discover the value themselves. No quotas. No compliance dashboards. Just access, training, and room to experiment.

The teams that figured it out saw 20x productivity gains. Not 20%. Twenty times. The ones that didn't were still filing the same reports the same way.

AI adoption is a culture problem, not a technology problem. Incentive alignment and autonomy beat mandates every time.

.27

Building the muscle

100

Product Managers

600

Engineers

40-50

Designers

Weekly AI upskilling sessions where teams share what they have built, what worked, what failed. Not workshops with slides. Live demos of actual workflows being transformed.

The knowledge compounds. Each team's discovery becomes institutional knowledge. Each breakthrough gets shared, adapted, and improved across the organization.

Tribal knowledge compounds. The team that starts learning today builds an insurmountable lead over the one that waits for a mandate.

.28
The Learning

The most valuable managers will be the ones who figure out how to diffuse AI throughout the company. Not the ones who use it personally, but the ones who make everyone around them better with it.

The new management imperative

.29

Act VI

The New Playbook

.30

Designing for agents

For decades, managers designed workflows for humans. Define a process, assign roles, set checkpoints, measure output. The new discipline is doing the same thing, but for agents.

Consider a product manager's workflow: gather customer feedback, synthesize insights, write a spec, prioritize features, coordinate with engineering. Every single one of those steps can now be performed by an agent with the right instructions and context.

The manager's role shifts from doing the work to architecting the system that does the work. You become the orchestrator, not the operator.

Replace labor with compute. Design workflows for agents the way you would design them for a team of brilliant, tireless interns.

.31

What every manager needs

01

Prompting

The ability to instruct AI precisely. Not vaguely, not hopefully, but with the specificity and structure of a well-written brief. Clear context, clear constraints, clear success criteria. This is the new literacy.

02

Evaluations

The ability to measure whether AI is performing. Build rubrics. Design test cases. Create feedback loops. If you cannot evaluate the output, you cannot improve it. This is the new quality assurance.

.32

Everyone is becoming a generalist

The old career advice was specialize. Find your niche. Become the world's expert in one narrow domain. AI inverts that logic. When any specialist skill can be accessed through a prompt, the premium shifts to the person who can combine skills across domains.

A designer who can code. A strategist who can build. A marketer who can analyze data, write copy, design the landing page, and deploy it. The boundaries between roles are dissolving, and the people who thrive will be the ones who refuse to stay in a single lane.

Sam Altman has predicted the one-person billion-dollar startup. It sounds hyperbolic until you realize the tools already exist to make it possible. The only missing ingredient is the person who knows how to use all of them.

Breadth is the new depth. The polymath, not the specialist, is the archetype of the AI era.

.33

Four actions, starting tonight

01

Build an AI Budget

Allocate Rs 2,000-5,000/month to AI tools. ChatGPT Pro, Claude, Cursor, Replit. This is not a luxury. It is an investment in remaining employable.

02

Build AI Fluency

Spend 30 minutes every day using AI for real tasks. Not tutorials. Not courses. Actual work. Write a report. Analyze data. Build a prototype. Learning happens by doing.

03

Think in Workflows

Pick any process you do repeatedly. Map each step. Ask: can an agent do this? Build the automation. Start with one workflow this week.

04

Become a Generalist

Learn to code (even a little). Learn to design (even roughly). Learn to market. The more skills you can orchestrate through AI, the more valuable you become.

.34

So let me ask again.
How many of you feel anxious?

Go build something.

The best way to understand the future is to create it.

Press F for fullscreen · Arrow keys or spacebar to navigate