AI Agent for 7-Platform Quick Commerce Intelligence

Industry

FMCG (Coffee, Snacks, Beverages, Personal Care)

Geography

Pan-India — 50+ cities, 15,000+ dark stores

Platforms Covered

Blinkit, Zepto, Swiggy Instamart, BigBasket, Amazon Now, Flipkart Minutes, JioMart

Data Coverage

Real-time pricing, stock levels, offers, discounts, delivery ETA, ad placements

Refresh Frequency

10-minute cycle on price/stock; daily on assortment

Delivery

REST API + Real-time dashboard + Slack/Email alerts

Client Overview

The client is a leading Indian FMCG company with significant presence across coffee, snacks, beverages, and personal-care categories. The brand sells through traditional retail, modern trade, and increasingly through India's booming quick commerce channel — which now drives 25-40% of urban FMCG revenue depending on category.

With 7 major quick commerce platforms competing aggressively across Indian metros, manual monitoring had become impossible. Stockouts on a competitor platform would trigger demand spikes on others within hours. Promotional pricing changes would ripple across the category in minutes. Ad campaigns underperforming on one platform would silently bleed budget while another platform showed strong ROAS. The client needed a real-time intelligence agent — not weekly reports.

Why Quick Commerce Demands a Real-Time Agent

Quick commerce moves faster than any retail format in India. Pricing changes hourly. Stock turns over multiple times a day. Dark stores activate and deactivate based on demand. A weekly PowerPoint dashboard is essentially a museum exhibit by the time it lands in inboxes.

Business Challenges

Before partnering with Actowiz, the client faced four interconnected operational gaps:

Challenge #1 — Real-Time Out-of-Stock Blindness

With over 15,000 dark stores across 7 platforms, knowing which SKU was OOS in which store at any moment was impossible manually. Stockouts on a competitor brand triggered demand the client couldn't capture; stockouts on the client's own SKUs went undetected for hours, costing direct sales.

Challenge #2 — Reactive, Not Predictive, Pricing

Competitor pricing changes were discovered after they had already shifted market share. The client repriced reactively — typically 24-48 hours behind. In a category where price elasticity peaks during festive and promotional windows, this lag translated directly into lost revenue.

Challenge #3 — Ad Spend Bleeding on Underperforming Platforms

Ad campaigns ran across all 7 platforms simultaneously, but performance varied dramatically by platform, city, and time of day. Without consolidated real-time ROAS visibility, underperforming ad sets continued spending budget for days before manual reviews caught them.

Challenge #4 — Fragmented Data Across 7 Platforms

Each platform had its own UI, refresh patterns, and dark-store structure. The client's team was juggling 7 dashboards, 7 reporting cycles, and 7 different ways of measuring 'pricing' — making cross-platform decisions slow and error-prone.

Project Objectives

Together with Actowiz Solutions, the client defined five measurable objectives:

• Detect out-of-stock events on client and competitor SKUs within 15 minutes across all 7 platforms
• Track competitor pricing in real-time with automatic anomaly detection
• Surface underperforming ad sets within hours, not days
• Deliver a single unified intelligence layer replacing 7 separate dashboards
• Enable autonomous AI-driven recommendations for pricing and ad-spend reallocation

Actowiz Solutions Approach

Actowiz built a 5-stage AI agent pipeline running on a continuous 10-minute cycle:

• CAPTURE
  Multi-platform crawl across 7 Q-commerce + 15K dark stores
• NORMALISE
  Unified SKU taxonomy across platforms
• DETECT
  ML-based OOS, price anomalies, ad signals
• DECIDE
  AI agent generates pricing & ad recommendations
• ALERT
  Real-time Slack/email + dashboard + API

Stage 1 — Hyperlocal Multi-Platform Capture

Actowiz built dedicated crawlers for each of the 7 platforms, simulating customer pin codes across 50+ Indian cities and 15,000+ dark stores. Residential proxy infrastructure ensured sustained capture without disruption. Browser automation handled JavaScript-heavy Q-commerce frontends, while anti-bot defences were navigated through human-like behavioural patterns.

Stage 2 — Unified SKU Taxonomy

Each platform had its own SKU naming, pack-size conventions, and category structure. Actowiz built a canonical taxonomy mapping every SKU across all 7 platforms to a single master ID — so that 'Continental Espresso Coffee Powder 200g' on Blinkit, 'Continental Espresso 200gm' on Zepto, and 'Continental Coffee Espresso (200g)' on Instamart all resolved to one canonical SKU. This made true cross-platform comparison possible.

Stage 3 — ML-Based Detection Engine

Three ML models ran continuously: (a) an OOS classifier detecting stockouts within 10 minutes; (b) a price-anomaly detector flagging unusual competitor moves against historical baseline; (c) an ad-performance scorer ranking ad sets by ROAS in real time.

Stage 4 — Autonomous Recommendation Agent

An LLM-powered agent consumed detection outputs and generated specific, actionable recommendations: 'Reduce price on SKU-X in Bangalore Blinkit by ₹4 to match competitor'; 'Pause ad set 12 on Zepto — ROAS down 38% in 4 hours'; 'Increase stock allocation to JioMart Mumbai dark stores — demand spike detected'.

Stage 5 — Real-Time Alert & Delivery Layer

Alerts flowed to Slack channels, email digests, and a real-time dashboard. A REST API exposed all data and recommendations for integration into the client's pricing engine and ad platforms.

Operational Wins

• OOS detection time: from 9 hours (manual reporting) to 12 minutes (AI agent)
• Pricing decision lag: from 24-48 hours to under 30 minutes for 90% of cases
• Ad spend efficiency: 42% ROAS improvement on Q-commerce ad budget
• Team time saved: 28 hours/week previously spent on manual cross-platform monitoring, now eliminated
• Replaced 7 separate platform dashboards with one unified intelligence layer

Client Testimonial

"Quick commerce moves faster than any retail channel we've ever competed in. Before Actowiz, we were always two steps behind — finding out about a stockout or a competitor price move after it had already cost us. The AI agent changed that fundamentally. Now we're responding in minutes, not days. The ₹14 crore uplift in six months is real money — but the strategic shift, from reactive to predictive, is worth even more."

— Head of Digital Commerce, Leading Indian FMCG Brand

Conclusion

Indian quick commerce is the fastest-moving retail format in the country — and arguably the world. With 7 major platforms competing across 15,000+ dark stores, traditional reporting cycles simply cannot keep pace. Stockouts, price changes, and ad performance shifts measured in hours, not days, demand intelligence measured in minutes, not weeks.

Actowiz Solutions delivered an AI agent that closed exactly that gap — capturing real-time multi-platform data, detecting events through ML, generating specific actionable recommendations through an LLM-powered agent layer, and delivering it all through alerts and APIs the client's teams could act on immediately. The result: ₹14 crore of measurable revenue uplift in 6 months, a 76% faster OOS response, and a 42% improvement in ad ROAS.

For Indian FMCG brands operating in quick commerce, the question is no longer whether to monitor the channel in real time, but how. The brands building real-time AI intelligence today are pulling away from those still on weekly dashboards — and the gap is widening fast.