As CDO of a multi-vertical conglomerate, how do you balance the need for centralised digital strategy with the autonomy each business unit requires?
At SGG our businesses span aviation, retail, FMCG, education, and manufacturing. Each operates under entirely different operating models, customer cycles, and unit economics. So, our approach is structural: centralise the digital foundation and decentralise the business innovation that sits on top of it.
At the group level we own cybersecurity, cloud governance, identity, data standards, ERP excellence, and AI governance. These are the layers where fragmentation creates risk and where scale creates leverage. At the unit level, each business retains the freedom to shape its own customer experience, operational design, and vertical-specific systems. Star Air’s PSS modernization looks nothing like Star LocalMart’s planogram intelligence, and it shouldn’t.
OneSGG is the connective tissue: independent businesses, shared intelligence, common data backbone.
Parag Bhagwat, Group Chief Digital Officer, Sanjay Ghodawat Group (SGG)
What is your framework for prioritising which AI initiatives get funding and which get shelved?
We evaluate AI initiatives against one question. Will this produce a measurable operational outcome within two quarters, and can the underlying data sustain it at scale? We greenlit production vision models for operational verification across multiple business lines, including retail compliance and solar panel quality. The pattern was consistent across both: visible manual cost in the P&L, environments controlled enough to make the model reliable, and clean reference data. Inspection cycles reduced materially.
We deferred a customer-facing chatbot pilot in another vertical because the underlying CRM data was not yet clean enough. Even a frontier model would have behaved unpredictably, and adoption would have failed on the data layer regardless of the model layer.
What separates AI initiatives that scale from the ones that quietly stall comes down to honesty about data readiness, not enthusiasm about models.
What were the biggest technology fragmentation challenges you inherited when you joined SGG, and how did the ‘OneSGG’ vision emerge from those?
The clearest fragmentation pattern when I joined was that each vertical had built its own version of nearly everything. Separate productivity stacks, multiple ERP instances, inconsistent identity, and parallel data definitions for the same operational concept. Each was justified in isolation. Aggregated across the group, the cost was visible, no cross-business visibility, no reusable AI substrate, and governance gaps in places no single business unit could see.
OneSGG emerged from that diagnosis. We approached transformation in two layers. The foundation first, covering cybersecurity, identity, productivity, cloud governance, and a unified data backbone. Then applied AI built on top of it. SGG One AI, our internal orchestration platform, sits on that backbone and gives every business unit access to the same models, governance, and workflow primitives.
How do you decide which technology to build in-house versus buy versus partner for across such a diverse portfolio?
The decision is rarely abstract. It depends on whether the capability creates differentiation for the group, how fast we need it in production, and whether mature platforms already exist.
For standard enterprise capabilities like productivity AI, common cloud services, and mainstream ERP modules, buying mature platforms is the rational default. Rebuilding them internally is a vanity tax.
For capabilities that touch our operational core or our AI substrate, we build. SGG One AI is the clearest example. No off-the-shelf product gave us the chaining, governance, and workflow assignment we needed across five verticals with different data shapes. So we are building it.
For specialised domains where deep vertical expertise compresses timelines materially, we partner. We do so selectively, with co-creation built into the contract.
Standardise where you can. Customise where it matters. Partner where the learning curve is not yours to climb.
With India’s DPDP Act coming into force, how are you building privacy by design into SGG’s digital architecture rather than retrofitting it?
The DPDP Act’s real shift is that consent, purpose limitation, and auditability move from being a compliance discussion to being an engineering decision. Once those become engineering decisions, they have to live in the foundation, not in a later compliance layer.
At SGG we are treating identity and access as the first lever. We are consolidating identity across business units, enforcing role-based access at the platform layer, and centralising audit trails. Fragmented identity is the single largest privacy debt most Indian conglomerates carry, and the one that becomes most expensive when DPDP enforcement matures.
The harder problem is data classification across business units that were never built to talk to each other. That is the work that defines the next two years. Without trustworthy data foundations, AI delivers exposure faster than it delivers value.
With AI models commoditising rapidly, where do you believe India’s sustainable competitive advantage in enterprise AI will come from over the next three years?
Models are commoditising. The competitive advantage is moving elsewhere, towards the systems and operational data that turn a generic frontier model into enterprise outcomes.
India’s edge over the next three years will not be engineering talent. That part has already been globalised. The edge will be operational data depth and applied-AI craft.
Indian enterprises run at extreme volumes, across airlines, retail chains, FMCG distribution, and financial services, under cost pressure most Western competitors do not face. The operational data generated is unusually dense, which is exactly what domain-specific AI needs. Combined with a tolerance for shipping AI into production while it is still learning, this creates conditions where applied AI compounds faster here than in markets where the bar for “ready” is higher.
The foundational models may come from global labs, but the operational systems that convert them into business outcomes will increasingly be built here in India.
Enjoyed this interview? Now imagine yours. Write to:
editor@thefoundermedia.com
About The Author
