Is AI Rewiring Enterprise Procurement — Who Really Wins? | 202X Vision
Host: Vishwendra Verma, Founder, GrowthSutra
Expert Panelists:
Siva Rao — Former Circle CEO at Indus Towers, (with strategic supply chain experience at Tata Steel, TCS, and Airtel)
Dharmender Kapoor— AI, Digital Transformation & Business Advisor; Former CEO, MindSprint and Birlasoft
Segment 1: The Invisible Leaks – Why Procurement Keeps Breaking in the Same Places
Q. Why is procurement suddenly getting so much attention from CEOs compared to "hero" functions like sales?
Siva Rao : Historically, the sales function is considered the hero function in the industry. When a CEO wants to increase profits by 10%, the immediate thought is to drive more sales. However, increasing sales is highly complex—it requires time, marketing spend, and market readiness.
In contrast, achieving that same 10% profit increase can be done much more simplistically by reducing procurement costs by just 5% to 10%. Procurement is an in-house function where vendors are within your control. Smarter procurement yields a direct, high-impact path to profitability that CEOs are finally recognizing as AI arrives
Q.Where do small and mid-size enterprises (SMEs) lose the most value before a single rupee is spent?
Siva Rao:Value leaks typically occur long before a supplier enters the picture, beginning right at demand origination and specification. There are three to four distinct pain points where value is lost:
Spend Visibility: Without clear visibility, organizations cannot segregate spending or leverage volume effectively.
Specification Clarity: A lack of item-level specifications prevents proper benchmarking and cost analysis.
Bill of Materials (BOM): If the BOM is poorly designed or unmanaged, procurement teams lack accurate benchmarks, such as a "should-be" cost, which leads to operational failure
Segment 2: The Unfinished Revolution – Where AI Re-Wires vs. Exposes What’s Broken
Q.Technology has promised to fix procurement for decades—from MRP and ERP to cloud P2P. What pattern keeps repeating that causes these transformations to stall?
Dharmender Kapoor :The pattern is repeating because organizations continue to make the exact same mistake for the last 30 years: disconnecting technology from the underlying process and data. Whether you have multiple fragmented systems or one monolithic ERP, having technology is not a metric for success. If your business processes aren't accurately mapped to the software, you create operational holes.
Q.How must AI architectures be fundamentally different to break this cycle?
Dharmender Kapoor :Traditional ERPs were built strictly for standardization and acting as systems of record. But modern business demands systems built for change. Sales, marketing, and production plans will always alter dynamically.
AI must act as a predictive system of action sitting on top of standardization. It must look at how the supply chain behaves and make simultaneous, real-time adjustments on the fly. However, if the foundational master data isn't standardized first, the AI will fail or hallucinate.
Segment 3: Capturing the Value – Can AI Agents Operate Autonomously?
Q.Can AI agents truly run sourcing events, analyze trade-offs, and execute from PR to PO autonomously?
Dharmender Kapoor :Yes, but we must understand the difference between analytics and autonomous decision-making. Predictive analytics solved the problem of "what to buy and when" years ago. The true evolution of an AI agent is its ability to take autonomous action based on guardrails.
For example, when a client requests a quote, you might validate the AI agent's response the first ten times. Once it consistently meets your operational guardrails, you let it take autonomous action without human intervention.
Siva Rao :Automation itself isn't new; even 20 years ago at TCS, we designed a system for a global chemical giant where if an incoming invoice matched the PO value within a threshold, the payment processed automatically. Today's agentic AI makes this much easier to deploy, provided your technical specifications, cost benchmarks, and ERP integrations are perfectly aligned.
Q. Will tech-deficient small suppliers be squeezed out as large enterprises race ahead with AI procurement?
Dharmender Kapoor :While it is a challenge due to messy historical data, it is actually a massive level playing field. Right now, nobody is a flawless expert at AI. Large companies have deep pockets, but that doesn't guarantee success. Smaller companies are nimble and can leapfrog legacy tech. Look at how quick-commerce startups like Blinkit disrupted massive giants like Amazon by rewiring last-mile supply chains. SMEs can use this shift to create true differentiation.
Siva Rao :Furthermore, as a small supplier, you don't necessarily need complex AI to sell to an enterprise—you just need to be AI-compatible. Ensure you are an approved vendor, structure your quotes so they are machine-readable, and align your data with their digital infrastructure so their AI automatically pulls your competitive bid.
Audience Q&A — AI agents, autonomy, and small-supplier squeeze
Q.Can AI agents autonomously analyze tradeoffs, run sourcing events, and cover PR→PO and vendor comparisons?
Dharmender kapoor: Be careful: predictive analytics for ‘what/when/how much to buy’ existed years ago. The AI agent leap is decision autonomy—evaluating options, selecting the best path, and acting within guardrails. Maturity path matters: validate early, then grant autonomy for bounded decisions (e.g., quoting within thresholds).
Siva Rao : Autonomy is possible when processes and masters are tight. Example: standardized bearing specs + approved supplier pool + price benchmarks + ERP-supplier integration = agent can evaluate 15 bids and place the PO within rules. Invoice auto-pay within ±5% of PO was live at a global chemical major back in 2007—today’s AI can extend that logic across more workflows.
Q: Small suppliers face a double squeeze—low tech maturity and messy data—while enterprise buyers adopt AI. Do they get left behind?
Dharmender kapoor : It’s a challenge—and a level playing field. No one has ‘mastered’ AI yet. Smaller firms can be more nimble and leapfrog, as seen in e-commerce and 10-minute delivery disruption. Distinguish with speed, clarity, and tight collaboration.
Siva Roy : To sell into large buyers, you don’t need your own AI to start. Become an approved supplier, submit structured prices aligned to the buyer’s ERP/AI portals, and ensure digital invoicing and performance transparency. To manufacture efficiently, yes—implement a light ERP, clean masters, and then add AI to optimize quality and cost.
Ownership, visibility, and why AI pilots die
Q: Between CPO, CFO, and CTO—who owns success? And what most often kills AI before production?
Siva Roy : The root cause isn’t titles—it’s the lack of a reliable digital backbone. Without ERP adoption, clean masters, and integration of shadow data (emails/Excels) into the system of record, AI has no footing.
Dharmender Kapoor : Two non-negotiables: ownership and measurement. Assign process and activity owners, and define KPIs upfront. Then build the core: visibility—of processes, exceptions, data quality, materials, spend, and actions. Without a shared ‘you are here’ map, teams point fingers and programs stall.
India advantage: turning rails into results
Q: India has structural rails (GST trails, mandated supplier digitization, UPI). PWC India estimates $135–150B AI value for MSME manufacturing by 2035, but maturity remains “middle.” What must change?
Siva Roy :
Re-engineer PR→PO→GRN end-to-end—truly digitize the first mile (demand/specs) and the last mile (GRN, 3-way/4-way match).
Build and maintain the core procurement masters and keep them current.
Institutionalize spend visibility with standardized tax/cost structures and benchmarking.
Dharmendar Kapoor :
Install a real digital backbone before scaling agents.
Acknowledge the new bottleneck: coding is cheaper; validation, reviews, data readiness, and governance are the constraints.
Think platforms, not just services: India can move up the stack with sector-specific procurement automation and agentic workflows.
Key Takeaways
To successfully rewire enterprise procurement via AI, organizations must focus on five non-negotiable checkpoints:
Fix the First Mile First: Resolve ambiguities in demand origination and product specifications before buying an expensive platform.
Clean Data Over Use Cases: The bottleneck is almost never the AI tool; it is the fractured data, shadow spreadsheets, and broken operating models underneath.
Appoint a Single Owner: AI doesn’t resolve process ambiguity; it amplifies it. Appoint one clear owner (CPO, CFO, or CTO) with a single set of metrics.
Measure Time to Production, Not Pilot: Implement a strict 90-day gate. If an AI pilot cannot show a clear pathway to live deployment within 90 days, kill the budget and reallocate it.
Treat AI Compatibility as an Asset: If you are a buyer, score vendors on their machine-readable data and digital invoicing. If you are a supplier in 2026, building clean data trails is no longer optional—it's a survival requirement.
Watch The Replay
This Q&A is based on GrowthSutra's "Is AI Rewiring Enterprise Procurement — Who Really Wins? | 202X Vision" session.
The full replay is available on GrowthSutra's LinkedIn and YouTube channel .
NEXT
The next session in the 202X Vision series is scheduled on 18 June 2026 — "Everything is Compliant. Is That Enough for the Future of Deals? | 202X Vision"
Register Here