Unified Commerce & AI ft. Jigar Dafda, CPTO at Fynd
For the last decade, e-commerce innovation has been incremental. Better storefronts. Faster checkout. More channels.
AI has snapped that curve.
In this episode of SaaS Sessions, Jigar Dafda, Chief Technology & Product Officer at Fynd, breaks down what’s actually changing in Indian e-commerce—and what most founders are still missing.
This isn’t about hype. It’s about structural shifts.
From Storefronts to Systems: How Fynd Evolved with Commerce
Fynd didn’t start as a “unified commerce platform.”
It became one—organically.
Started with storefronts
Expanded to offline retail
Integrated marketplaces (Amazon, Flipkart)
Unified inventory, OMS, WMS, ERP into one stack
Key insight:
E-commerce products that survive a decade don’t predict the future—they evolve with customer behavior.
That evolution now has a forcing function: AI.
AI Was Always There. What Changed Is Access.
Machine learning has powered personalization, forecasting, and recommendations for years.
The difference now?
LLMs removed the gatekeepers.
Before:
Data lived in silos
Insights required analysts, SQL, dashboards
Decisions were slow and indirect
Now:
Anyone can ask questions in plain English
Analysis happens in real time
Intelligence is embedded into workflows
“The data scientist layer is going invisible.”
AI didn’t create new capabilities.
It democratized existing ones.
The 3 Biggest AI Impact Areas in E-commerce
1. Buying Surfaces Are Shifting (Conversational Commerce)
Websites won’t disappear—but they won’t dominate.
Conversations become the primary interface
Discovery, comparison, and checkout happen via chat
Payments are embedded inside AI interfaces
Merchants still own fulfillment and data
SEO is giving way to GEO
Search Engine Optimization → Generative Engine Optimization
If your product isn’t understandable by AI agents, it won’t be recommended.
2. Hyper-Personalization Becomes Default (Not a Feature)
AI + CDPs (Customer Data Platforms) = precision at scale.
What changes:
Personalized pricing, delivery options, content
Unified online + offline customer views
Contextual recommendations, not generic ones
What doesn’t change:
Data is still the oil
Governance and consent still matter
Uncomfortable truth:
If 100 users visit your site and don’t buy, you’re burning infrastructure—not building a brand.
Hyper-personalization is no longer optional. It’s a conversion survival strategy.
3. Customer Support Is the First Function to Go Autonomous
Customer service is AI’s easiest win.
Why?
60–80% of queries are repetitive
Status checks don’t need humans
Multilingual support scales instantly
Modern AI agents:
Work 24/7
Pull live data from OMS, WMS, TMS
Respond in local languages
Escalate only edge cases to humans
Result:
Smaller teams. Faster resolution. Lower costs. Higher trust.
Dynamic Pricing: Powerful, Profitable—and Problematic
AI enables real-time pricing based on:
Demand
Location
Device
Urgency
Behavior history
From a business lens: effective.
From a consumer lens: uncomfortable.
Jigar’s stance is pragmatic:
Personalization must be opt-out capable
Data control must sit with the user
Regulation will catch up (slowly, but inevitably)
AI will push personalization to the edge.
Regulation will pull it back.
The tension between the two defines the next decade.
The Back Office Is Where AI Quietly Wins
The real transformation isn’t customer-facing. It’s operational.
AI is reshaping:
Inventory planning
Demand forecasting
Campaign scheduling
Category management
Martech workflows
Previously:
Insights were retrospective
Decisions were manual
Execution lagged
Now:
Systems recommend actions
Pricing updates happen automatically
Forecasts adjust in real time
Autonomous commerce is emerging.
Humans don’t tell systems what to do.
Systems tell humans what should be done.
3 Predictions for Indian E-commerce
1. Conversational Commerce Goes Mainstream
Chat becomes the default buying interface.
2. Autonomous Back Offices Become Table Stakes
AI agents manage pricing, supply, and campaigns.
3. Work Itself Gets Redefined
Fewer specialized roles
More AI-augmented generalists
Faster building with smaller teams
“What took three engineers earlier, one person can now do in weeks.”
Advice for Founders: Build for What’s Coming, Not What Exists
The biggest risk today isn’t building the wrong feature.
It’s building for a present that won’t last.
Tech cycles are compressing
Interfaces are changing
Roles are dissolving and merging
Founders who win will:
Track emerging behaviors, not just competitors
Design systems that adapt
Assume today’s workflows will be obsolete
Final Thought
AI isn’t a feature you add to e-commerce.
It’s a force that reshapes how buying, selling, and operating work—end to end.
Those who treat it as a layer will fall behind.
Those who rebuild around it will define the next decade.
