Seven modules. One promise: simple.

01Ingest & Speed-to-Lead

Webhook receivers for Meta Lead Ads, Google Lead Forms, portal CSV and Click-to-WhatsApp. Every lead is deduplicated, language-inferred and durably stored before we acknowledge the webhook, then a personalised, RERA-compliant WhatsApp message goes out within 30 seconds at p95. Zero lead loss, 90-day replay.

• Idempotent ingest, five duplicate deliveries produce one lead, one message
• Language inference: ad locale → city → name heuristic
• First-touch latency recorded on every single lead

02Qualification & Site-Visit Booking

Native WhatsApp Flows collect budget, configuration, locality, timeline and financing without ever leaving the chat. When buyers type free-text instead, including code-mixed Hinglish, LLM extraction fills the same structured schema. Then slot inventory, channel-partner routing and reminder cadences take the lead to the project gate.

• Explainable lead scores, every score shows its contributing factors
• T-24h and T-2h reminders drive no-shows below 22%
• Bidirectional sync with Sell.do, Zoho, Salesforce, Leadrat and webhooks

03The Closed Loop

Qualified-lead, site-visit-booked, site-visit-attended and booking events fire back to Meta Conversions API and Google Offline Conversion Import, deduplicated against your pixel, with value attached. Meta stops optimising for form-tappers and starts finding buyers. The dashboard reports CPSV by campaign, creative, micro-market and channel partner.

• Event match quality ≥ 7.0, verified in Meta Events Manager
• Spend reconciliation within ±2% of platform reporting
• First-touch and position-based attribution models

04Consented AI Voice

When a WhatsApp lead says “call me,” an AI agent calls back within 60 seconds, in their language, on the correct regulated number series, opening with an unskippable AI disclosure. It only asserts facts from the project fact base, and hands off to a human on request, on frustration, or on any pricing negotiation.

• Sub-second conversational turns tuned for Indian 4G telephony
• Hindi, Marathi, Tamil, Telugu, Bengali + Hinglish code-mix
• The dial path is gated on a consent record, not a config flag

05Dead-Lead Revival & CP Intelligence

Every developer sits on a dormant lead database with no tool for it. Dhanur AI runs re-consent flows (mandatory under DPDP), segments by staleness and score, and re-engages inside a fresh WhatsApp window. Meanwhile the CP ranker routes new leads to the partners who actually convert, and gives CPs their own WhatsApp bot for claims and visit logging.

• Industry benchmark: 8-9% of dead leads revived, 5-6% of total bookings
• CP ranking on assigned→visit→booking conversion and response latency
• Inventory-aware messaging surfaces slow-moving units to high-fit leads

06RERA Data Graph

A normalised, continuously refreshed graph of RERA projects and agents across Maharashtra, Karnataka, Telangana, Tamil Nadu, UP and Gujarat, roughly 85% of registered projects. It powers project verification, fact-base seeding for grounded AI conversations, and agent onboarding checks.

• Change-detection crawl with 24-hour freshness targets
• Entity resolution across promoter name variants
• Scraped contact details are never used for outbound, enforced at the schema level

07Society Bulk Services

Housing societies and RWAs register once, then run pooled service drives: AC servicing, painting, pest control, deep cleaning and more, booked across dozens of flats at once. Vendors compete for the pooled job, discounts unlock as more residents join, and the same WhatsApp engine handles opt-ins, scheduling and reminders.

• Tiered group discounts that grow with participation, up to 35% at 100+ flats
• Committee dashboard: quotes, vendor ratings, completion photos, one consolidated bill
• Constant resident traffic that compounds into the network for resale and rental demand