CX

Omnichannel CX Intelligence Platform

Multi-Tenant Customer Experience Intelligence & AI Routing Grounded in Enterprise SOP Repositories

G
Active User Connecting...
Event Flow:
Google Maps Review
→
GCS Media Store
→
Cloud Pub/Sub
→
Vertex AI Gemini 3.8 Flash + SOPs
→
BigQuery Data Lake
→
Human-In-The-Loop Approval

Google Maps Review Simulator

Exact GBP Schema
Quick Scenario Fill (Executive Demo):
★ ★ ★ ★ ★

Drag and drop photo or click to browse

Supports ATM error screens, receipts, queue chits (JPEG/PNG)

Live CX Intelligence & Triage Console

Real-time stream from BigQuery & Pub/Sub grounded in enterprise SOP repositories

Loading recent reviews from BigQuery...

Executive CX Analytics & Root Cause Matrix

Direct BigQuery aggregation across enterprise locations with zero simulation

Total Reviews Analyzed
0
BigQuery Data Lake: rhb_review_intelligence
Average Star Rating
0.0 ★
Across all nationwide enterprise locations
Critical & High Risk Escalations
0
Immediate SOP action required
Human Approval Rate
0%
Human-In-The-Loop governed replies

Customer Sentiment Distribution

Positive Sentiment 0 (0%)
Neutral Sentiment 0 (0%)
Negative Sentiment 0 (0%)

Risk Classification Breakdown

Critical Risk (Fraud / Swallowed Card) 0
High Risk (SLA Breach / Surge) 0
Medium / Low Risk 0

Top Diagnosed Root Causes (BigQuery Grouping)

Clustered by frequency
Rank Diagnosed Root Cause Category Primary Risk Level Occurrences
No records found in BigQuery. Submit reviews to see Pareto distribution.

AI Response Quality & Closed-Loop Feedback Analytics

Tracking human modifications, edit drivers, and continuous SOP/prompt optimization in BigQuery
RLHF / Active Learning Loop Active
Direct AI Acceptance Rate
100%
Approved As-Is with zero edits
Human Edit Frequency
0%
0 of 0 modified by agent
Average Edit Distance
0 ch
Levenshtein character delta
Flywheel Health Status
CALIBRATED
Continuous Feedback Stream

Top Reasons Behind Human Modifications

All responses currently approved as-is. Modify an AI draft to record feedback reasons.

Closed-Loop Prompt & SOP Enhancement Insights

Baseline Stability: AI draft acceptance is high. When customer service agents modify responses, this engine clusters the edit rationales to recommend exact phrasing and SOP clause enhancements.

Recent Human-in-the-Loop Modification Audit Log

Review ID Customer & Branch Primary Modification Reason Agent Feedback Note Edit Delta Approval Time
No modified responses recorded yet. Responses approved without edits are tracked as 100% As-Is.

Live BigQuery SQL Query Inspector (Demonstrating Production Enterprise Architecture)

-- 1. Review Triage & Risk Matrix Aggregation SELECT COUNT(*) AS total_reviews, ROUND(COALESCE(AVG(rating_value), 0), 2) AS avg_rating, COUNTIF(sentiment_label = 'POSITIVE') AS positive_count, COUNTIF(sentiment_label = 'NEUTRAL') AS neutral_count, COUNTIF(sentiment_label = 'NEGATIVE') AS negative_count, COUNTIF(escalation_level = 'CRITICAL') AS critical_risk_count, COUNTIF(escalation_level = 'HIGH') AS high_risk_count, COUNTIF(reply_status = 'APPROVED') AS approved_count FROM `rhb-cx-intelligence.rhb_review_intelligence.reviews`; -- 2. Closed-Loop AI Quality & Human Modification Analytics SELECT COUNT(*) AS total_approvals, COUNTIF(is_modified IS NOT TRUE) AS approved_as_is, COUNTIF(is_modified IS TRUE) AS modified_by_human, ROUND(COUNTIF(is_modified IS NOT TRUE) * 100.0 / COUNT(*), 1) AS ai_acceptance_rate, modification_reason, COUNT(*) AS modification_frequency, ROUND(AVG(edit_distance), 1) AS avg_edit_distance FROM `rhb-cx-intelligence.rhb_review_intelligence.review_approvals` GROUP BY modification_reason;

Standard Operating Procedures (SOPs)

Indexed in BigQuery Catalog & Cloud Storage Document Store for Gemini Cognitive Grounding

SOP Code Title Department Target SLA Actions

Enterprise Architecture & BigQuery Quota Analysis

1. Does storing each review into BigQuery upon receipt make sense?

YES, absolutely. This is the recommended Google Cloud enterprise architecture for Customer Experience & Reputation Intelligence:

  • Permanent Data Retention: Google Business Profile UI limits historical review access (often truncated to 18 months). Storing in BigQuery ensures permanent, unalterable historical telemetry.
  • Cross-Corpus Omnichannel Joins: BigQuery enables enterprises to cross-join public Google Maps reviews with internal transactional logs, ticketing systems, and campaign attribution.
  • Multi-Tenant Isolation: Data rows are partitioned by ingestion date and clustered by tenant_id, guaranteeing strict data segregation per enterprise while running on shared Google Cloud serverless infrastructure.

2. Will we hit any BigQuery quota issues?

NO. BigQuery default quotas are orders of magnitude larger than any retail or commercial review volume:

  • BigQuery Storage Write API Quota: Standard default quota allows 100,000+ rows per second and up to 2 GB per second per project.
  • Enterprise Ingestion Scale: Nationwide retail chains and banks ingest 100 to 1,000 reviews per day (< 1 review per second peak). This consumes less than 0.001% of BigQuery's ingestion quota.

3. Event-Driven Architecture Topology

The solution utilizes Google Cloud native serverless components for high resilience and zero operational overhead:

  1. Google Business Profile Webhook / Pub/Sub: When a customer submits a review on Google Maps, GBP publishes a notification event to Google Cloud Pub/Sub topic gbp-review-events.
  2. Serverless Cloud Run Ingestion: Subscriber ingests the event, handles any attached photos/receipts via Cloud Storage in asia-southeast1.
  3. Vertex AI Gemini 3.8 Flash Grounding: Gemini evaluates the review + photo against the tenant's SOP repository. It generates internal operational steps (with exact SOP citations) and crafts a brand-compliant customer reply.
  4. Human-In-The-Loop Approval Gate: The operations manager reviews and approves the draft before it is posted back to Google Maps via the GBP API.