Case Study Real Estate & Sales SaaS

ClosIQ — AI lead scoring, straight inside WhatsApp

Sales agents in Southeast Asia were losing deals to slow replies. We built a platform that answers every WhatsApp inquiry in 4 seconds and tells agents exactly which leads to chase first.

4 secauto-reply to every inquiry — down from 2.5 hours
+52%more deals closed per agent, per month
92%of hot leads answered — up from 45%
4 monthsfrom concept to live beta with a 6-person team
The Challenge

Leads go cold in hours, not days

Property agents, car dealers, and insurance brokers across Singapore, Malaysia, and Vietnam field dozens of WhatsApp inquiries a day — with no way to tell a serious buyer from a browser. Typical response time was 2–4 hours. By then, the prospect had messaged three competitors.

No prioritizationAgents spent hours on low-intent prospects while warm leads sat unanswered.
Forgotten follow-upsHigh-value leads slipped through the cracks with no reminders or escalation.
Zero team visibilityEvery agent ran their own ad-hoc system — managers had no data at all.
The Solution

Every message scored in real time

ClosIQ plugs into the WhatsApp Business API — no new app for agents to learn. Every incoming message flows through an AI scoring pipeline and lands on the right agent's dashboard in seconds.

01Message arrivesWhatsApp webhook received, signature validated
02Cache checkRedis hot-path returns recent scores in under a millisecond
03AI scoringML model trained on 50k+ conversations rates intent 0–1
04ClassifyHot / Warm / Cold, with per-agency thresholds
05Route & alertHot leads push to the agent instantly; warm leads batch twice daily
Lead lifecycle guardrailsDay 2 with no follow-up? The agent gets an at-risk alert. Day 5? Auto-escalated to the team lead. Nothing falls through.
Tuned per agencyAggressive teams flag leads at a 0.4 score; conservative ones at 0.7. Personalization was built in early — and it compounds retention.
Built to scaleStateless workers on Kubernetes hold the 4-second guarantee at 1,000 messages a minute, with blue-green zero-downtime deploys.
Technology

The stack behind the 4-second guarantee

MessagingWhatsApp Business APIAuthenticated channel — no new app for agents
BackendNode.js + ExpressAsync I/O for real-time scoring throughput
DataPostgreSQL + RedisDurable audit trail plus sub-millisecond hot cache
AI / MLCustom scoring engineRules + XGBoost model trained on agent history
InfrastructureAWS + KubernetesMulti-region: Singapore, Malaysia, Vietnam next
DeliveryGitHub Actions + ArgoCDAutomated tests, blue-green deploys, 99.5% SLA
Results

50 agents, 3 agencies, one beta

MetricBeforeAfterChange
Average response time2.5 hours4 seconds−99.7%
Hot-lead response rate45%92%+47pp
Deals closed per agent / month2.13.2+52%
Lead follow-up rate60%88%+28pp
Leads handled per agent / day40802× capacity

Auto-reply SLA held at 99.2% compliance through beta; classification accuracy 76% with per-agency calibration ongoing.

Two Products, Two Architectures

The right architecture for the problem

ClosIQ

B2B SaaS — real-time, multi-region

Heavy backend: Node.js, PostgreSQL, ML pipeline on AWS & Kubernetes

6-person team, 4 months to beta

PDFCatalyst

Privacy-first productivity tool — everything runs in the browser

Zero backend: WebAssembly + static CDN, ~$50/month to run

4-person team, 10 weeks to launch

Read the PDFCatalyst case study

Building something with real-time demands?

Tell us what you're building. We'll reply within 24 hours with a tailored proposal and team recommendation.

Start Your Project