Transforming Business Through AI

Contents

Executive Summary

This panel discussion explores enterprise AI adoption in India and globally, emphasizing that the critical differentiator is not frontier model development but applied AI execution and outcome-driven value creation. The speakers argue that India's competitive advantage lies in infusing AI across economic sectors, building AI-native enterprises, and creating digital public infrastructure (akin to UPI for payments) rather than competing with US/China on foundational models.

Key Takeaways

The Moment is Now: Enterprise adoption is accelerating (20% say AI is top priority, 53% say top three). If you're not moving, you risk disruption or obsolescence. Real, measurable ROI stories exist today.
Shift from Technology Selling to Outcome Ownership: Providers must move from selling software (per-seat, fixed pricing) to owning and sharing in measurable business outcomes. This is the fundamental business model reset.
India's Path Forward is Applied AI + Digital Public Infrastructure: Rather than competing on foundational models, India should invest in (a) infusing AI across economic sectors, (b) upskilling the IT services workforce, and (c) building a unified AI interface (UAI) as digital public infrastructure. This creates defensible, global-scale opportunities.
Execution > Intelligence: The technology (LLMs) is solved. The bottleneck is moving from 70% pilot accuracy to production scale while managing data, legacy systems, organizational change, and measurement of real impact.
Reframe the Narrative: Stop debating labor displacement in isolation. Focus instead on whether AI enables order-of-magnitude improvements in capital efficiency, revenue, or customer outcomes. India's population scale is a reinforcement learning advantage, not just a cost-cutting tool.

Key Topics Covered

Enterprise AI Adoption Patterns: Global and India-specific adoption rates, the "breakaway 10%," and the transition from pilot purgatory to production scale
Pricing & ROI Models: Evolution from seat-based to outcome-based pricing; challenges in measuring and communicating AI ROI
Job Displacement & Human Impact: Labor transition narratives, retraining, compassionate change management, and examples of net job creation vs. role transformation
India's Role in Global AI: Applied AI vs. frontier models; comparative advantage; building indigenous foundational models and digital public infrastructure
Enterprise Transformation Mechanics: Data integration, legacy system modernization, agentic workflows, and capital/materials cost savings (not just labor)
Funding & Portfolio Strategy: How venture investors evaluate AI opportunities; focus on outcome-driven, full-stack providers vs. tech-only solutions
Unified AI Interface (UAI): India AI Mission initiative to create open digital public infrastructure for AI, analogous to UPI

Key Points & Insights

The "Breakaway 10%": Among Indian enterprises, only 10% are significantly accelerating AI adoption. These companies think differently—they chase outcomes, not promises, and recognize enterprise AI as fundamentally different from consumer AI (e.g., ChatGPT as a generalist vs. specialized enterprise agents).
Outcome-Based Pricing is Becoming Non-Negotiable: Traditional SaaS models (per-seat, time-based) are obsolete. Forward-thinking providers are shifting to risk-sharing models where pricing is tied directly to measurable P&L impact (cost savings, revenue generation, or basis points of realized value). Token-based pricing is transitional and imperfect.
ROI is Broader Than Labor Savings: Most discussions fixate on job displacement, but major cost impacts come from elsewhere:
- Free cash flow improvements ($200M+)
- Inventory reduction ($400M+)
- Software/IT spend reduction ($250M via ticket automation)
- Material/inventory optimization in capital-intensive industries
- Labor is only one lever in a larger system.
Data Architecture & Tech Debt Are Unglamorous But Essential: Real enterprise AI success requires solving boring, expensive foundational problems: unified data architecture, consolidating legacy systems, data pipelines. This has no near-term ROI but is a prerequisite for scale.
Pilot Purgatory is a Major Barrier: Many companies achieve 70-80% accuracy in sandbox pilots using general-purpose LLMs but fail to translate that into production-scale ROI. The hardest part of AI is execution, not intelligence. Companies are skeptical of moving from pilot to scale.
Order-of-Magnitude Improvements, Not Incremental: Leading companies target 5x–80x improvements (autonomous/AI-assisted workflows covering up to 80% of work), not 5% efficiency gains. This mindset shift—from cost-cutting to business reinvention—separates winners from also-rans.
India's Comparative Advantage is Applied AI + Talent + DPI:
- Not frontier models (US/China focus)
- Applied AI infusion across sectors (manufacturing, finance, healthcare, agriculture)
- Deep talent pools for AI transformation services (upskilling IT services workforce)
- Unique digital public infrastructure assets (1.3B population data, 10-12B UPI transactions/month, Indic languages)
- Cost structure and outcome focus allow beating Western competitors domestically and winning emerging markets
Population as an Asset, Not a Liability: India should reframe its 1.3B population not as a labor-cost advantage (which AI erodes) but as a reinforcement learning advantage—human feedback at scale is essential for advancing toward AGI and building better foundational models.
Change Management is Non-Negotiable: The human element—retraining, reskilling, compassionate dialogue with affected teams—is essential but often overlooked. Real examples: UK hospital saved 4,000 lives via process automation; unionized European teams transitioned when errors (not jobs) were eliminated. Each scenario requires hundreds of conversations.
The "Outflank, Don't Catch Up" Moment: India can skip outdated infrastructure (e.g., landline → mobile) and leap directly to AI-native business models. With no legacy trillion-dollar investment constraints, India can move faster than incumbents and dominate emerging markets (Europe, Africa, Southeast Asia, Middle East) where talent/models/applications are scarce.

Notable Quotes or Statements

Mahir Shukla (Automation Anywhere): "Enterprise AI is very different than how most of us think about it on ChatGPT. Think of ChatGPT as your English teacher. When you give an English teacher a math problem, a physics problem, an engineering problem, of course it has hallucinations."
Mahir Shukla: "Chase outcome, not promise." (Key differentiator of the breakaway 10%)
Karthik Ready (Bloom Ventures): "It's not a typical SaaS—'here's I'm going to charge me $1,000 per seat.' Fundamentally we've seen companies 10x in one year because they're directly pointing towards outcomes."
Shravan Faludu (Avatar.ai): "The hardest part of AI is not intelligence... it's execution."
Mahir Shukla: "I don't want to talk about labor. I want to save in material. I want to save in cash because your 90% savings will come out of that [in capital-intensive businesses]."
Jinme Sharma (InfoEdge Ventures): "There will be companies that are NGMI [not gonna make it]. The real conversation to have is not what model can I use... The real conversation is: How are my data pipelines working? How do I build a unified data architecture?"
Mahir Shukla (on India's role): "India doesn't have to focus on frontier models. Everything we wanted to do, if there was no more investment in frontier models, we can do 80% of everything we want to do... It's about infusing AI into every sector of the economy."
Mahir Shukla: "You don't want to catch up, you want to outflank [other economies]."
Shravan Faludu: "AI is an existential question. Can you really afford to not become AI-first?"
Karthik Ready (on India's opportunity): "There is a UPI-like moment waiting to happen [in AI]... that suddenly catapults India to a very different opportunity set than playing piggyback on what the US is developing for us."

Speakers & Organizations Mentioned

Entity	Role	Organization
Arpanched	Moderator	Bain & Company
Mahir Shukla	Founder & CEO	Automation Anywhere (RPA & agentic process automation)
Karthik Ready	Founder & Managing Partner	Bloom Ventures (VC firm, $850M AUM)
Shravan Faludu	Founder & CEO	Avatar.ai (applied AI platform)
Jinme Sharma	Venture Capitalist, Lead Partner in AI	InfoEdge Ventures
—	Multiple portfolio companies	Bloom, Grey Orange Robotics, Purple, Gani.ai, Exotel, Squad, Confido Healthcare (US), various others
—	—	India AI Mission (building Unified AI Interface)
—	—	UC Berkeley (knowledge partner for Avatar.ai)
—	—	Sequoia Capital, Tiger Global (Avatar.ai investors)
—	—	Seven of top 10 Indian banks (Automation Anywhere clients)
—	—	5,000+ global customers, 100+ GCCs, 100+ universities (Automation Anywhere)

Technical Concepts & Resources

RPA (Robotic Process Automation) & Agentic Process Automation: Automation Anywhere's focus; evolution toward AI agents handling complex, autonomous workflows
LLM (Large Language Model): General-purpose models (ChatGPT, Claude, etc.) vs. specialized enterprise models
Foundation Models / Foundational Models: Frontier models (developed by OpenAI, Anthropic, etc.) vs. indigenous/distilled models for specific use cases
Model Distillation: Technique to create smaller, efficient models from larger ones (mentioned as key technique India can leverage)
Reinforcement Learning from Human Feedback (RLHF): Path toward AGI; India's population scale as an advantage for feedback generation at scale
Unified AI Interface (UAI): India AI Mission initiative; digital public infrastructure modeled on UPI for payments
DPI (Digital Public Infrastructure): Examples cited: UPI (11-12B transactions/month), Aadhaar, 1.3B population data
Indic Languages & Indian Voice Accents: Localization challenges that foundational models must address
Legacy System Modernization & Data Integration: Unglamorous but essential foundational work (consolidating 45+ legacy systems, unified data architecture, tech debt payoff)
Agentic / Autonomous Workflows: Processes where AI handles 80% autonomously or with AI assistance; moving from generalist to specialist models
Token-Based Pricing: Current standard in LLM pricing; acknowledged as transitional and imperfect for enterprise customers

Data & Metrics Referenced

Metric	Value	Source/Context
Global enterprises rating AI as top priority	20%	Bain & Company survey (updated quarterly)
Global enterprises rating AI as top-3 priority	53%	Bain & Company survey
Projected increase in top-3 prioritization	~75%	Bain & Company (next survey prediction)
Automation Anywhere customers globally	5,000+	—
AI agents in Automation Anywhere production	420 million	—
AI agents in India production	Tens of millions	Subset of 420M
GCCs using Automation Anywhere	100+	—
Top Indian banks using Automation Anywhere	7 of top 10	—
Universities training on Automation Anywhere AI	100	—
Students trained via universities	~1 million	—
Nonprofit organizations using AI agents	50+ (e.g., Akshay Patra)	—
Bloom Ventures AUM	$850 million	Across multiple funds
Avatar.ai funding	$55 million	Sequoia Capital, Tiger Global
Avatar.ai patents	10	—
Breakaway enterprises (India)	~10%	Mahir Shukla's observation
Free cash flow improvement (example)	$200 million	Automation Anywhere customer case
Inventory reduction (example)	$400 million	Materials company example
IT software spend reduction (example)	$250 million	Ticketing system automation (84% autonomous)
Capital savings (example)	20,000 crores	Large Indian capital-intensive company
Ticket automation rate	84% autonomously	Large enterprise IT customer
Customer service automation rate (target)	Up to 80% autonomous/AI-assisted	"Autonomous Enterprise" model
India UPI transactions/month	10-12 billion	DPI asset
India population	1.3 billion	DPI asset for RLHF advantage
OpenAI revenue (calendar year 2024)	$18-20 billion	—
Anthropic revenue (calendar year 2024)	Similar to OpenAI	—
Company growth rates (OpenAI, Anthropic, Palantir)	Rapid, 18-20% annual	Enterprise penetration accelerating

Context & Setting

Venue: AI Summit (appears to be India-focused or multi-region)
Panel Format: 4 panelists + 1 moderator; dialogue-oriented rather than Q&A
Audience: Enterprise executives, CTOs, founders, investors, IT services leaders (referenced in commentary)
Timing: References to "last month," "yesterday," and "next session" suggest this is a multi-day conference
Related Session: Bharat Jen presentation follows