Render isometrik smart city catchment dengan marker geospatial di atas toko, jalan, dan kendaraan
Special ProjectAgentic Geospatial IntelligenceReliance Retail

RetailVista.

Layer geospatial intelligence enterprise Reliance Retail. Memprediksi potensi sebuah catchment. 3 track, 24 kategori data, scoring 10 dimensi pada shared H3 hex index. 68 live opportunities.

Live - Internal Track - Opportunity Explorer on grocery NSO leadsPilot - Google Track - MVP demoed end-to-end on Mumbai catchmentsBuilding - JioGIS data unlock, Foundation modelRoadmap - 1000-city scale
StatusInternal Track live01-May-2026

3

track berjalan

68

live opportunities

10

dimensi LI

24

kategori data

Retail Vista in action

Status

Yang berjalan. Yang sedang dibangun. Yang berikutnya.

Tiga track, per 01-May-2026.

Live

Internal Track - UCP / JioMart

Opportunity Explorer.

Dibangun in-house di atas UCP design language. Berbasis data produksi BAU JioMart yang diagregasi pada level hex. Natural-language workspace menilai catchment, menandai risiko cannibalisation, dan mengusulkan intervention zone.

Pilot

Google Track - Joint MVP

Agentic Site Intelligence.

Joint build dengan Google di Vertex AI, BigQuery, Street View, dan Maps. Agent cannibalisation, feasibility, dan site-selection divalidasi end-to-end pada catchment Mumbai live. Fynd memegang Product, requirement, evaluasi CUJ, dan feedback kualitas output.

Building

JioGIS Track

Data unlock + Foundation model.

Mendorong unlock layer JioGIS dan dataset Reliance Retail. Pengembangan foundation model dimulai untuk spatial reasoning core. Kentrix MMR procurement menjadi shared enrichment layer yang memberi makan ketiga track setelah secured.

Opportunity

Modul yang berjalan hari ini.

Internal Track surface berjalan di UCP design language. Setiap modul adalah Opportunity Explorer flow yang membawa catchment dari scored signal menuju pipeline decision. Live per 01-May-2026 pada real grocery NSO leads, diagregasi di data JioMart.

ModuleStatusCapability
Home - Command Center
Live
Ringkasan KPI, Opportunity Intelligence (Comp. Gaps, Whitespace, Active Leads, Approved), Top Opportunities, Leads, Alerts Requiring Attention, Store Rollout Velocity vs target
Explorer - Discovery + Analysis
Live
Daftar opportunity yang bisa difilter (Site / LI / Whitespace / Competitor sub-scores), Google Maps tersinkron, Quick Summary, Full Analysis dengan 10 dimensi LI, Discover full-screen map, Pipeline Kanban (New -> Shortlisted -> Site Visit -> Approved)
Workspace - AI Co-Pilot
Live
Multi-turn chat untuk feasibility, catchment, drive-time, brand-network analysis; output AI terstruktur (demographics, accessibility, competitor landscape, brand assets); toggle Google Maps + MapLibre engine; POI dataset dan clustering layers
Users - Access Management
Live
User table, roles, status, modules, last login, invite flow, permission management
Data & Scoring layer
Live
10 dimensi LI (GDP, Demographics, Accessibility, Property Rates, Wealth, Footfall, Spending, Order Data, Building Density, Competition), 4 sub-score menjadi composite 0-100, Confidence dan Priority labels (High / Medium / Low)

Yang dilihat operator.

Enam surface dari Internal Track. Setiap catchment dinilai di 10 dimensi Location-Intelligence; sub-score menjadi composite 0-100 dengan Confidence dan Priority labels.

RetailVista Home - Command Center dengan Opportunity Intelligence dan Top Opportunities

Home - Command Center: 68 opportunities, 0 approved, 0 shortlisted. Opportunity Intelligence panel, Top Opportunities, Leads feed, Alerts, dan Store Rollout Velocity vs Q2 FY26 target.

Explorer dengan scored opportunity list dan Google Maps tersinkron

Explorer - scored list + map: opportunity list yang bisa difilter dan tersinkron dengan Google Maps. Category dan vertical filters. Quick Summary popup pada map markers.

Explorer Full Analysis dengan 10 dimensi LI dan composite score breakdown

Full Analysis - 10 dimensi LI: Goregaon East, Score 69, Medium Confidence. Composite 62/100 lintas GDP, Demographics, Accessibility, Property Rates, Wealth, Footfall, Spending, Order Data, Building Density, Competition.

Explorer Pipeline Kanban dengan kolom New, Shortlisted, Site Visit, Approved, Dismissed

Pipeline - Kanban: pipeline lima stage (New, Shortlisted, Site Visit, Approved, Dismissed). Decision system-of-record yang menghubungkan setiap opportunity ke outcome dunia nyata.

Workspace AI co-pilot dengan feasibility analysis output untuk Andheri

Workspace - AI Co-Pilot: multi-turn chat dengan ranked hex IDs, rwi_mean, Total POIs, epoch_mean, height_mean, strategic rationale, dan contrast analysis.

Users access management table

Users - access management: member table, role, status, modules, last login. Lima active members di tim build Internal Track hari ini.

Path to L4

Bagaimana RetailVista menjadi L4 agentic.

Saat ini berada di L3. Tiga track paralel menutup gap menuju L4.

TrackStateTarget rungYang dibuka track ini
Internal - UCP / JioMart
Live
L4
Opportunity Explorer, 68 scored leads, agentic decisioning end-to-end.
Google - Joint MVP
Pilot
L4
Agent cannibalisation, feasibility, dan site-selection di Vertex AI untuk catchment Mumbai.
JioGIS
Building
L4
Spatial reasoning core, Foundation model, memberi makan ketiga track setelah secured.
Arsitektur

Dari raw signals ke activated decision.

Lima layer: Sources, Spatial Aggregation, GIS Visualisation, Agentic Orchestration, Activation. Agent membaca sinyal live, menjalankan guarded playbook, dan menulis outcome kembali untuk continuous learning.

Live

Layer 01 - Sources

24 kategori.

UCP, JioGIS, Retail store master, Customer locations/demographics/behaviour/intent, Government data (Census, NCRB, RBI, RERA, NFHS), third-party POI (Kentrix).

Live

Layer 02 - Aggregation

H3 hex index.

Hex-first aggregation, tanpa PII. Event dipetakan ke H3 cells; hanya aggregated metrics per hex yang sampai ke model. Satu spatial truth dengan banyak zoom.

Pilot

Layer 03 - GIS visualisation

Maps + heat overlays.

Pewarnaan score per-hex pada catchment Mumbai hari ini. Toggle Google Maps + MapLibre engine. Street View untuk last-mile. Pan-India per-hex Attractability heat sedang dibangun.

Mixed

Layer 04 - Agentic orchestration

Agent skills.

Enam skill dipetakan: 3 Live (New Store Opening, Catchment Analysis, Dark-store Drive Time), 1 Building (Customer Sentiment), 2 Roadmap (Pricing Promotion, Land Parcel, Transport Optimisation).

Building

Layer 05 - Activation

Hand-off surfaces.

Output ke ALP, Granary, routing JioMart, dan workflow brand-team. New stores, Pricing, SCM Optimisation, Customer Listening menjadi activation cells dari cornerstone deck.

Arsitektur RetailVista dari Sources hingga Use Cases

Diagram arsitektur: lima stage dari source ke outcome. Sources melalui Ingest, Transform/Process/Serve, Analyze, hingga Use Cases.

Data

Setiap sinyal yang mendarat pada hex.

Shared data inventory yang memberi makan ketiga track: Reliance proprietary, Jio, Government, dan third-party POI.

CategorySourceRefreshGranularityVariables
01 - Land base & public dataset
JioGIS
As-is
Lat / Long
36 states, 105M buildings, 137M households, 21.7M POIs, 0.6M villages, 25.6K cities, 3.8M km
02 - Buildings master (Residential / Commercial)
JioGIS
As-is
Lat / Long
1.3M km Fiber, 0.3M eNodeB OnAir
03 - Owned Retail facilities
Retail store master
Daily
Lat / Long
40,000+
04 - Customer locations
UCP - Jio
Real-time
Lat / Long
500M+
05 - Customer demographics
UCP - Jio
Real-time
Lat / Long
20+
06 - Customer digital behaviour
UCP - Jio device
Daily
Lat / Long
10+
07 - Customer purchase intent
UCP transactional
Real-time
Lat / Long
150+
08 - Customer interests & propensities
UCP - Jio - Media inferred
Daily
Lat / Long
700+
09 - Civic Infrastructure
Govt (Mission Antodaya, ODP)
As-is
200m - 1km
240+
10 - Commercial - Services
POI (Kentrix)
30 days
Lat / Long
38
11 - Commercial Retail
POI (Kentrix)
30 days
Lat / Long
110+
12 - Crime statistics
Govt (NCRB)
As-is
500m
110+
13 - Demography
Govt (SECC, Census) + GeoIQ
180 days
200m - 2000m
230+
14 - Environment
Govt (Aridity, IMD)
As-is
200m - 500m
2
15 - Finance
POI (RBI data)
30 days
Lat / Long
23
16 - GeoIQ Indices
GeoIQ engineered
180 days
500m
16
17 - Geographical
GeoIQ engineered
180 days
Region
4
18 - Healthcare
Govt (NFHS, Census, SECC) + POI
As-is / 30 days
200m - 1000m
140+
19 - Infrastructure
POI (Kentrix), OSM, Public
30 days / 1 yr
Lat/Long, 200m - 1000m
80+
20 - Leisure & Hospitality
POI (Kentrix)
30 days
Lat / Long
24
21 - MSME
Govt (third-party)
3 months
500m
100+
22 - Mobility & Footfall
Third-party
30 days
Hex 8
20
23 - Real Estate
Govt (RERA), Public listings
30 / 90 days
500m, Lat/Long
3+
24 - Socio-economic
Govt (SECC, Census) + GeoIQ
180 days
200m - 2000m
900+
RetailVista Platform Layers five-tier diagram

Platform Layers v0.1: Sources, Spatial Aggregation, GIS Visualisation, Agentic Orchestration, dan Activation.

Visualisasi H3 hex Mumbai Attractability scoring dengan popup Ghatkopar

H3 hex visualisation, Mumbai: Attractability score per-hex (0-100) di seluruh MMR. Sample: Ghatkopar 53/100, Jio Penetration 43%, Competitors 4, Cannibalisation Risk 9%.

Vision

Arah RetailVista berikutnya.

Backbone geospatial pan-India di 1000 kota, berbobot pada agentic stack Google, dibangun di sekitar per-household Digital Twins dan network reference untuk setiap operasi Reliance.

Roadmap

01 - Scale

1000 kota.

Coverage spatial pan-India di level kota. Pilot hari ini berpusat di Mumbai pada Internal Track dan Google MVP. Scale bergantung pada unlock data JioGIS, Kentrix MMR enrichment, dan maturasi foundation model.

Pilot

02 - Primary surface

Google sebagai pan-India agentic surface.

Bobot engineering dan partnership bergerak ke Google joint track. Agent cannibalisation, feasibility, dan site-selection sudah divalidasi end-to-end pada catchment Mumbai.

Building

03 - Foundation

Full-stack geospatial foundation.

Platform foundation 12-layer yang mencakup sources, aggregation, GIS, agentic orchestration, dan activation. Arsitektur 5-layer saat ini adalah v0.1.

Building

04 - Operating model

Joint execution, dedicated org.

Organisasi RetailVista dedicated yang beroperasi lintas Reliance dan Fynd. Integrasi engineering dengan Google masuk ke satu execution plan.

Roadmap

05 - Customer Digital Twin

Per-customer Digital Twin, CDAP-native.

Setiap pelanggan dirender sebagai Digital Twin konsumsi berbasis hex: wallet, channel mix, household, fibre, mobile. Mendorong personalisasi offer saat decision time.

Roadmap

06 - Network reference

Satu GIS layer untuk setiap operasi Reliance.

Coverage jaringan 5G + 4G, dark stores, RIL Neighbourhood stores, dan Enterprise Premise Connectivity ditampilkan dalam satu GIS reference.

Roadmap

07 - Household identification

100M households identified, path to 180M owned.

Resolusi street-map all-India untuk mengidentifikasi 100M household dari Broadband dan Air Fibre footprint, dengan jalur menuju 150-180M owned-home relationships.

Roadmap

Setiap pertanyaan bisnis menjadi agent skill.

Use case dipetakan ke Activation Layer dan Agentic Orchestration cells dari cornerstone architecture. Set pertama Live; sisanya Building atau Roadmap.

Use caseStatusYang dilakukan
New Store Opening
Live
Skor setiap catchment; pre-score sebelum ground visit; cannibalisation, demand, dan feasibility dicek sebelum approval.
Catchment Analysis
Live
Scoring catchment level hex lintas format. Workspace AI co-pilot menjawab feasibility, catchment, drive-time, dan brand-network questions.
Dark-store Drive Time
Live
Drive-time isochrones untuk q-commerce. Set use case pertama dari cornerstone executive summary.
Customer Sentiment Analysis
Building
Data Customer Listening diagregasi secara agentic ke level hex.
Pricing & Promotion
Roadmap
Elastisitas harga dan efektivitas promosi level hex.
Customer Listening
Roadmap
Overlay spatial dari voice-of-customer signals.
Transport & SCM Optimisation
Roadmap
Optimasi route transport terhadap spatial backbone.
Land Parcel Availability
Roadmap
Surfacing land parcel untuk Strip Mall dan ekspansi large-format.
Inventory Availability
Roadmap
Presence inventaris level hex vs demand signal.
Total Addressable Market (TAM)
Roadmap
TAM per format pada resolusi hex.
AI summary
RetailVista adalah layer geospatial intelligence enterprise Reliance Retail yang dibangun Fynd untuk memprediksi potensi sebuah catchment. Sinyal dari UCP, JioGIS, sumber government, dan third-party POI diagregasi ke shared H3 hex index di 24 kategori data, lalu dinilai di 10 dimensi Location-Intelligence menjadi composite score 0-100. Tiga track berjalan paralel: Internal Track live di production, Google Track piloting agentic site intelligence di Vertex AI, dan JioGIS track yang membuka data serta foundation model.

Dibangun untuk bisnis seperti Anda. Mari terhubung

  1. 1

    Isi formulir

    Bagikan informasi kontak Anda untuk memulai

  2. 2

    Bicara dengan pakar

    Anggota tim sales kami akan menghubungi Anda

Hubungi kami

Dengan mengirimkan formulir ini, Anda menyetujui Ketentuan Layanan dan Kebijakan Privasi kami.