Our client is a French real estate company that made a bold strategic pivot: rather than competing as another property listings platform, they repositioned as an AI-powered assistant that helps users find their ideal home through natural conversation on WhatsApp.

The concept is simple but technically demanding. Instead of browsing listings manually, buyers and renters describe what they're looking for in their own words, and an AI agent handles the search on their behalf. This product is offered as a white-label solution to B2B real estate agency partners, giving each agency their own branded, AI-driven client experience.

The platform was originally built on a pragmatic technology stack for an early-stage startup racing to market but unequipped for the new direction.

Delivering a scalable, multi-tenant AI agent product required rethinking the architecture from the ground up. The client partnered with us to make that transformation happen.

Real estate agencies have long relied on the same playbook: list properties, wait for inquiries, and manually match clients to listings. It works until the volume of clients grows faster than the team can handle it.

Our client saw this problem clearly and had an answer: an AI assistant that could handle property searches conversationally on WhatsApp – a platform their clients already use daily. Turning that vision into a white-label product was a different challenge entirely.

The biggest obstacle was the foundation, since the existing platform was a monolithic MVC architecture which made it nearly impossible to create separate, isolated environments for each agency partner. Onboarding a new agency meant a slow, manual process. Any update had to be rolled out individually, one instance at a time. The operational overhead was significant, and it only threatened to get worse as the partner network grew.

At the same time, the client needed to ensure that each agency's data and client interactions remained separate and secure. It was a non-negotiable requirement in a competitive, trust-sensitive industry like real estate.

The challenge, in short, was this: how do you build a scalable, white-label AI product on top of a system that wasn't designed to support one?

We adopted an agile delivery approach, prioritizing core infrastructure and building toward the full product vision in deliberate, well-defined stages.

Architecture was the most critical shift. We redesigned the backend to support a multi-tenant model, giving each agency partner a fully isolated, independent environment while operating on a shared underlying platform. As a result, we eliminated the manual, instance-by-instance deployment process that had been creating operational bottlenecks, replacing it with a standardized onboarding pipeline and centralized feature rollouts across the entire partner network.

With the foundation in place, we developed the centerpiece of the product: an AI-powered WhatsApp assistant integrated via the WhatsApp Business API. The solution allows homebuyers and renters to describe their requirements in natural language and receive relevant property matches without app downloads or complex search interfaces. The system automatically routes each interaction to the appropriate agency, ensuring every partner maintains a distinct, branded client experience.

To handle the volume of concurrent conversations and webhook events the system requires, we architected the backend on Azure Function Apps – a cloud infrastructure purpose-built for on-demand scalability. We also implemented asynchronous event processing and batch database operations to sustain performance under growing load.

On the client-facing side, a Next.js frontend rebuild delivered measurable improvements in page speed and search engine visibility, strengthening the platform's organic presence while the B2B product scaled.

Third-party integrations completed the solution: Stripe for subscription and payment management between the platform and its agency partners, and SendGrid for automated client communications.