Fundle.ai: Retail Loyalty Automation with AI Agents for Indian Malls

Defining Retail Loyalty Automation with AI Agents

Retail loyalty automation with AI agents is distinct from marketing automation and from traditional loyalty management systems. Marketing automation — think WebEngage, MoEngage, or Xeno — schedules and sequences communications based on rules a human sets. A loyalty management system like Capillary or EasyRewardz manages points ledgers, tier thresholds, and redemption catalogues. Both are necessary. Neither is sufficient.

An AI agent, in the context of loyalty, is a goal-directed software entity that perceives inputs — transaction data, visit signals, app events, ad engagement, CRM history — reasons over them using a model, and takes actions autonomously to advance a defined objective: maximize repeat visit probability, maximize basket size at next visit, prevent churn, or upgrade a customer's RFM tier. The agent does not wait for a campaign brief. It runs continuously, re-evaluating every customer's state after every new signal.

Consider what this means practically. A customer walks into a mall anchor — say, a Reliance Trends at a Nexus mall — on a Tuesday afternoon, which is atypical for her. Her last three visits were Saturday mornings. The AI agent flags this as an anomalous visit, cross-references her points balance (she is 180 points short of a Silver tier upgrade), and triggers a real-time push notification: 'You're 180 points from Silver. Shop at Pantaloons today and get there.' The agent does not need a campaign calendar entry. It does not need a human to approve the communication. It acts within seconds of the visit event.

This is what separates agentic AI from automation. Automation executes instructions. Agents pursue goals. Fundle AI Agents are architected around this distinction — they are given business objectives (reduce churn in the 60-90 day lapsed segment, increase cross-brand visits per enrolled member) and autonomously select the right action, channel, timing, and offer to achieve those objectives. The agent layer sits above the engagement channel and the points engine, orchestrating both.

Tailoring Loyalty Automation Strategies for Indian Malls

India is not a monolith. A loyalty strategy that works for a 1.2-million-sq-ft super-regional mall in Mumbai will not translate directly to a community mall in Indore or a high-street format in Chandigarh. Several India-specific structural realities must shape how AI-driven loyalty automation is architected.

First, the tenant mix problem. Indian malls have a dramatically higher share of F&B and entertainment tenants than their Western counterparts — often 35-40% of gross leasable area. These categories generate high visit frequency but low basket values and thin margins. A loyalty agent optimizing purely for transaction value will under-invest in F&B anchors that are actually driving footfall for apparel and jewellery. The agent's reward function must balance category-level economics: a visit to Cafe Coffee Day that precedes a Tanishq purchase within the same 90-minute window is worth far more than its ₹320 bill suggests.

Second, the UPI identity layer. India's payment infrastructure has created a unique opportunity. UPI transactions carry a verified mobile number and, increasingly, a mapped bank account. For malls that have integrated with QR-based POS systems — whether on POSist, Petpooja, GoFrugal, or Wondersoft — there is a real-time transaction signal tied to a verified identity that Western loyalty operators can only dream of. Fundle Mall Loyalty is built to ingest this UPI signal natively, enabling identity resolution without requiring the customer to download an app or scan a loyalty card.

Third, the festival calendar is not optional — it is the business. Diwali, Eid, Navratri, Pongal, Christmas: Indian retail is intensely seasonal and regionally heterogeneous. An AI agent operating in a Chennai mall must weight Pongal differently from one in a Jaipur property. Fundle AI Agents are trained on India-specific retail seasonality data, enabling them to pre-position offers and tier incentives in the four-week build-up to a regional festival rather than reacting to the footfall spike when it arrives.

Fourth, the language and channel reality. India's loyalty communication stack is not WhatsApp-only or app-only. It is WhatsApp Business API, SMS, regional-language push notifications, and increasingly, QR codes at POS. A loyalty agent that cannot route a message in Kannada to a Bengaluru customer and in Hindi to a Lucknow customer will generate measurably lower open and redemption rates. Fundle's AI Workflow layer handles multi-language, multi-channel orchestration as a core function, not an add-on.

Integration with Mall Retail Media and CRM Systems

The integration question is where most loyalty transformation projects stall. A mall operator has 80-120 tenants. Each tenant may be running its own loyalty scheme — Manyavar has its own CRM, Apollo Pharmacy has its own app, Lifestyle has its own tier program. The mall management company wants a coalition layer on top. Between these two layers sits a rats' nest of POS systems, ERP connectors, and data consent complications.

Fundle AI Platform approaches this as a data mesh problem, not a data warehouse problem. Instead of requiring all tenant data to flow into a central lake before any intelligence can be applied, Fundle's integration architecture operates on event streams. A transaction event from a GoFrugal POS at a pharmacy counter, a visit event from a Wi-Fi probe at a mall entry gate, an ad impression event from a digital display in a food court — all of these are treated as first-class signals by the Fundle AI Workflow engine, processed in near real-time, and used to update each customer's loyalty state.

Fundle manages 3,759+ retail ad spaces driving loyalty and engagement automation — a number that reflects the scale at which the platform is already operating across mall and brand networks in India. This retail media integration is critical because it closes the attribution loop. When a customer sees a Tanishq offer on a mall digital display, clicks a WhatsApp CTA, and redeems within 72 hours, Fundle AI Agents can attribute that conversion with confidence and use it to train the next iteration of the offer model.

For enterprise CRM systems already deployed — Capillary, EasyRewardz, or custom-built stacks — Fundle operates as an intelligence layer rather than a replacement. The points ledger, the tier logic, and the customer master can stay where they are. Fundle's AI agents consume the data via API, apply agentic decisioning, and push actions back to the engagement channels. This hybrid architecture reduces implementation risk dramatically and typically allows a mall operator to go live with AI agent-driven campaigns within 8-12 weeks of contract signature, rather than the 12-18 month timelines that full platform migrations demand.

Measuring Impact: Increased Sales and Engagement

The measurement framework for AI-driven loyalty automation must be built before the first agent goes live, not retrofitted after the first quarter's results. This is not a data science nicety — it is a commercial necessity. Mall operators presenting to investors and brand tenants need clean attribution. CMOs at anchor brands like Lifestyle or Pantaloons need to see whether the coalition loyalty spend is generating incremental transactions or merely rewarding customers who would have visited anyway.

The primary metric for a mall loyalty program is incremental visit frequency: how many additional visits per enrolled member per quarter can be directly attributed to agent-triggered interventions, over and above the baseline visit rate of a statistically similar holdout group. Indian mall benchmarks put baseline visit frequency for enrolled members at 2.1-2.4 visits per month. Fundle's AI Agents consistently target a 0.4-0.6 incremental visit uplift — that is, 0.4 to 0.6 additional visits per member per month attributable to the program. Across a base of 1,50,000 enrolled members, that is 60,000-90,000 incremental mall visits per month. At an average spend of ₹1,800 per visit, that is ₹10.8 crore to ₹16.2 crore in monthly incremental tenant revenue.

Secondary metrics that matter for the CRM Head's dashboard: RFM tier migration rate (what percentage of the Promising segment upgraded to Loyal within 90 days), cross-brand visit rate (what percentage of members visited three or more distinct tenant categories in a 30-day window), and offer redemption velocity (the time between offer delivery and redemption, which is a proxy for offer relevance). A redemption velocity under 48 hours signals that the AI agent is getting offer timing and value right. A velocity over 7 days suggests the offer is too generic or the channel is wrong.

Competitors in the Indian loyalty automation space — Almonds.ai, Customer Capital, Antavo — offer campaign management and points management capabilities. The differentiator for platforms with true agentic AI is the shift from reporting on what happened to predicting what will happen and acting on it autonomously. When a Fundle AI Agent detects that a previously high-frequency customer has not visited in 38 days — still within the Needs Attention window but approaching the At-Risk threshold — it acts without waiting for a weekly segment refresh or a campaign brief. That proactive, autonomous decisioning is what drives the incremental visit numbers.

How Fundle solves this

Fundle was purpose-built for the Indian organized retail context — not adapted from a Western loyalty platform with India bolted on as a geography setting. The Fundle AI Platform integrates the four components that Indian mall operators need to run AI-driven loyalty at scale: an identity resolution engine built for UPI-era data, an agentic decisioning layer that pursues commercial outcomes rather than executing rules, a retail media integration layer that closes the attribution loop across 3,759+ managed ad spaces, and a multi-channel AI Workflow orchestrator that handles regional language, WhatsApp, SMS, and app push natively.

Fundle Mall Loyalty addresses the coalition complexity that defeats most mall loyalty programs. Rather than requiring all tenants to migrate to a single POS or CRM system, Fundle Brand Loyalty connectors integrate at the API and event-stream level with existing stacks — whether a tenant is running on Wondersoft, POSist, or a proprietary system. The points wallet is unified at the mall level; the intelligence is applied at the individual customer level; and the commercial outcomes are reported at the tenant level. This three-layer separation is what makes the platform deployable in a real Indian mall environment rather than a PowerPoint one.

Fundle AI Agents are the runtime execution layer. Each agent is configured with a goal, a set of allowed actions (offer types, channels, timing windows), and a set of guardrails (cost-per-redemption caps, frequency caps, category exclusions). The agents run continuously, re-evaluating every enrolled customer's state after every new signal. Fundle Agentic AI means that when a customer in the At-Risk segment makes an uncharacteristic Tuesday visit, the system does not wait for the next campaign cycle — it acts within minutes, selecting the optimal offer from the offer library, routing it through the optimal channel in the customer's preferred language, and logging the action for attribution.

Fundle AI Workflow is the orchestration backbone that coordinates actions across channels and across tenants. When an agent decides to send a cross-brand offer — visit Tanishq and earn triple points on your next Lenskart purchase — the Workflow layer handles the consent check, the tenant notification, the message delivery, the conversion tracking, and the points crediting, all in a single automated flow. Vineet Narang's founding vision for Fundle was precisely this: that India's retail operators should not have to choose between the richness of a coalition program and the precision of a personalized loyalty experience. Fundle delivers both, at the speed and scale that agentic AI makes possible.