NexaRealEstate
A concept AI co-pilot for German real estate agents. One intelligent system for leads, communication, and operations, designed and prototyped to explore how AI could replace the spreadsheet and WhatsApp stack agents rely on today.
NexaRealEstate is a concept product I built under NexaForgeAI to explore how an AI co-pilot could work for German real estate agents. It has no live users or business metrics. This case study covers the problem space, the design thinking, and the build.
The Problem
German real estate agents juggle fragmented workflows. Leads arrive from ImmoScout24, ImmoWelt, social media, and referrals. Agents track them across spreadsheets, sticky notes, and WhatsApp threads. Follow-ups slip through the cracks, and admin work eats into the time that should go to closing deals.
The opportunity I wanted to explore: a single tool built around the way German agents actually work, rather than another generic CRM.
Research & Discovery
Before designing a single screen, I studied the German real estate market. I mapped the daily agent workflow, looked at how tools like ImmoScout24 fit into it, and noted where the gaps appear. The core insight shaped everything that followed: agents do not need more features. They need fewer tools that work together.
I chose to design a German-first product rather than a translated American one. German agents use WhatsApp for business communication, rely on ImmoScout24 as their primary lead source, and need GDPR compliance in every workflow. These are not features you bolt on. They shape the entire product architecture.
What I Designed & Built
Intelligent Lead Hub
Centralized lead management with AI scoring that ranks prospects by budget, timeline, and fit. Automated nurturing sequences trigger follow-ups across email and WhatsApp, so a warm lead never goes cold.
Smart Communication Layer
A unified inbox pulls WhatsApp, email, and portal messages into one view. The AI drafts contextual responses in German and English, and viewing confirmations and scheduling happen automatically.
Operations Dashboard
A visual sales pipeline with drag-and-drop stages, conflict-free calendar integration, and real-time performance analytics. Agents and agency owners get full pipeline visibility in one place.
Build Stack
I built the working prototype with Next.js 14, TypeScript, and React 18. Supabase handles data, Clerk handles auth, and the AI features run on LLM APIs (OpenAI, Gemini, and Anthropic) with n8n for automation flows.
Interface Gallery
Key Decisions & Tradeoffs
- German-first UX. Workflows, terminology, and defaults are designed for the German market, not translated from an American product. The tradeoff is a narrower audience in exchange for a far better fit.
- GDPR by design. Consent management, data handling, and opt-in flows live in every touchpoint rather than being added at the end.
- AI as assistant, not replacement. The AI drafts, suggests, and automates, but the agent always has the final say. Trust matters in real estate, so I kept a human in the loop on anything client-facing.
- Built for non-technical users. Every interaction targets agents who live in WhatsApp, not developer tools.
Outcome & What's Next
The result is a fully interactive prototype, designed in Figma and built in Next.js. As a concept build it has no live users or business metrics. It served as a hands-on way to pressure-test the product idea: how one AI system could cut admin time, keep leads from leaking, and let small agencies scale without losing their personal touch.
Next, I would put the prototype in front of real agents to test the lead scoring logic and the AI response quality against live workflows, then decide which features earn their place in a v1.
View Live Prototype →