Silah tripled its qualified sales pipeline in one quarter by deploying an AI sales agent on WhatsApp

Silah handled 400+ WhatsApp inquiries daily across a portfolio of luxury and mid-market properties. Their six-person sales team was spending 60% of their time on pre-qualification: budget checks, location preferences, buying timelines. This happened before any real conversation started. Response times averaged three hours. In their market, that meant losing deals.

The sheer volume of inbound inquiries was both a blessing and a bottleneck. Silah's marketing team had built a strong digital presence, property listings on major portals, targeted social media campaigns, and a well-trafficked website, all funnelling inquiries into a single WhatsApp business number. But the sales process had not scaled to match. Each inquiry started with the same set of questions: What is your budget? Which areas are you considering? Are you looking to buy or rent? What is your timeline? Are you a first-time buyer or an investor? Do you need mortgage assistance?

Six agents rotated through the inbox in shifts, but response times still averaged three hours. During peak periods, typically Sunday evenings and weekday lunch hours, messages could sit unanswered for five or six hours. In Dubai's competitive real estate market, where a serious buyer might message five agencies simultaneously, a three-hour response time was effectively a lost lead. Competitors with faster response systems were winning the same prospects.

The pre-qualification process was also inconsistent. Each agent had their own style of questioning and their own threshold for what counted as a qualified lead. Some agents spent thirty minutes in a WhatsApp conversation only to discover the prospect's budget was well below the minimum for any available property. Others rushed through qualification and handed off leads that were not genuinely ready to view. There was no standardised handoff process, agents would verbally brief each other or drop a few notes in a shared spreadsheet, and context was regularly lost between the qualification conversation and the first viewing.

We built a WhatsApp AI agent on GPT-4 with a live knowledge base of Silah's full property inventory, pricing, and availability. The project began with a two-week design phase where we mapped the existing qualification workflow, interviewed all six agents to capture their best qualifying questions and objection-handling approaches, and analysed three months of WhatsApp conversation logs to identify the most common inquiry patterns and drop-off points.

The AI agent operates as the first point of contact on Silah's WhatsApp business number. When a new message arrives, the agent responds within seconds with a natural, conversational greeting and begins the qualification flow. The conversation feels organic rather than form-like, the agent asks about the prospect's requirements in a conversational sequence, adapts its questions based on previous answers, and handles follow-up questions about specific properties, neighbourhoods, or market conditions along the way.

The agent qualifies leads across eight criteria: budget range, preferred locations, property type (apartment, villa, townhouse), number of bedrooms, buying timeline, purchase purpose (end-user or investor), mortgage requirement, and nationality (relevant for ownership regulations in certain freehold zones). Each criterion feeds into a lead score that determines routing priority. The knowledge base behind the agent contains Silah's complete inventory, over 200 active listings with pricing, floor plans, availability status, developer details, handover dates, and payment plan structures. This data syncs every fifteen minutes from Silah's property management system via a REST API integration.

When the agent identifies a match between a prospect's requirements and available properties, it shares relevant listings with images, key details, and pricing directly in the WhatsApp conversation. If the prospect expresses interest in viewing, the agent checks agent availability via a Google Calendar integration and books the viewing, sending calendar invitations to both the prospect and the assigned agent. The handoff includes a structured brief, a summary of the qualification conversation, scored criteria, properties of interest, and any specific questions or concerns the prospect raised.

High-value leads, those with budgets above a defined threshold or matching specific premium inventory, trigger an immediate Slack notification to the senior sales team with a conversation summary, bypassing the standard queue. The system also handles common post-qualification tasks: sending location maps, sharing payment plan calculators, and following up with prospects who went quiet after initial qualification.

Response time dropped from 3 hours to under 30 seconds. The AI agent handles the initial greeting and begins qualification instantly, regardless of time of day or inquiry volume. This alone transformed Silah's competitive position, prospects who previously would have moved on to a faster-responding agency now received immediate, relevant engagement.

Qualified lead volume increased 3x. The improvement came from two sources. First, faster response times meant fewer leads were lost to competitors. Second, the AI agent's consistent, thorough qualification process captured leads that human agents had previously dismissed too quickly or failed to follow up with. The agent never skips a question, never forgets to ask about timeline, and never gets impatient with a prospect who takes hours to respond between messages, it picks up every conversation exactly where it left off.

Sales team time on pre-qualification dropped from 60% to under 10%. The six-person team was effectively operating as a three-person team before, because more than half their working hours went to repetitive qualifying conversations. After deployment, agents received only pre-qualified leads with full context briefs. They could review the qualification summary, look at the matched properties, and walk into a viewing or a call already knowing the prospect's budget, preferences, and timeline. Several agents reported that their first substantive conversation with a prospect now happened at the viewing itself, rather than over days of WhatsApp back-and-forth.

Qualified pipeline tripled in the first quarter post-deployment. Revenue impact followed, Silah closed more deals in the first quarter after launch than in the previous two quarters combined. The system also surfaced data insights that were previously invisible: peak inquiry hours, most-requested neighbourhoods, average budget bands by nationality, and common drop-off points in the qualification flow. Silah's marketing team now uses this data to optimise ad spend and listing placement. The AI agent currently handles over 85% of all initial inquiries autonomously, with human agents stepping in only for complex scenarios involving off-plan negotiations, bulk investor deals, or VIP clients.