eXp Realty Deploys AI Platform to Replace Millions in SaaS Contracts

Quick Brief

• The Shift: eXp Realty eliminated millions in annual SaaS costs by building custom software with Lovable, an AI development platform, replacing ChatGPT Enterprise, internal communications tools, and international website vendors.
• The Scale: The deployment spans 83,000 remote agents across 26 countries, with support tickets dropping 85% on international infrastructure and over 70% across internal tools.
• The Timeline: Non-technical teams built production-ready applications in hours one critical website launched in 6 hours to replace a failed vendor delivery.

eXp Realty, one of the world’s largest real estate brokerages with 83,000 agents operating entirely remotely, has terminated millions of dollars in SaaS contracts after deploying Lovable, an AI-powered software development platform. The company announced in January 2026 that it replaced ChatGPT Enterprise, eliminated per-seat licensing for internal communications, and rebuilt its international website infrastructure cutting annual costs by millions while improving performance across all deployments.

The Build-Versus-Buy Calculation That Changed

eXp’s SaaS expenditure scaled linearly with growth, creating unsustainable economics as the brokerage expanded internationally. Per-seat pricing models for agents, departments, and countries drove costs higher each quarter, while vendor delivery timelines stretched into months. Seth Siegler, Chief Innovation Officer at eXp, identified three critical pain points: ballooning per-seat costs, vendor velocity failures, and rigid one-size-fits-all platforms that forced workarounds instead of enabling workflows.

The breaking point occurred hours before a Peru country launch, when eXp’s external vendor failed to deliver a promised website. Sarah Hutchinson, VP of International Marketing, used Lovable to build the site in 6 hours a deployment that became the template for replatforming 26 country websites. The company had experimented with ChatGPT, Cursor, and OpenAI API tools, but Founder Glenn Sanford discovered Lovable in early 2025 and committed the organization to the platform.

Also Read Cisco Announces Silicon One G300: How 102.4 Tbps Switching Transforms AI Data Centers

Financial Impact Across Three Major Replacements

The international infrastructure rebuild delivered the largest immediate impact. eXp built a custom CMS for country sites, self-service agent microsite portal, and backend service normalizing data from multiple different CRMs into a single API. Canceling the vendor contract eliminated per-agent fees of $11/month, while site quality improvements reduced agent support requests by 85%.

AdwaitX Analysis: The No-Code-to-Full-Stack Shift

This deployment represents a significant shift in enterprise AI tooling strategy from consuming SaaS to building proprietary infrastructure using AI platforms that generate production code. Lovable differs from traditional no-code builders by producing full-stack codebases (React, Tailwind, Supabase) that sync to GitHub, not visual blocks. This architectural choice allows eXp to own and modify the underlying code, avoiding vendor lock-in while maintaining enterprise control.

The company empowered non-technical employees to build production software using Lovable and Supabase, according to Supabase case study documentation. Real estate agents with no coding background created functional applications during internal training sessions. Siegler stated that the barrier between conceptualizing and building software “finally dissolved”.

This approach challenges the SaaS-first procurement model that has dominated enterprise software for two decades. eXp now evaluates every vendor contract with the question: “Can we build this ourselves?” Siegler’s declaration “Every SaaS is on the chopping block” signals a strategic pivot that could reshape software procurement in distributed, digitally native organizations.

Technical Architecture: From Prompt to Production

eXp’s Lovable deployments follow a consistent stack: React frontends with Tailwind CSS, Vite build tooling, and Supabase for backend infrastructure (authentication, databases, storage). The platform accepts natural-language prompts and generates editable source code across the full application stack. For “The Hub” internal communications platform, Founder Glenn Sanford built the initial version himself to replace functionality lost when Meta sunset Facebook Workplace.

The system integrates with OpenAI and Google Gemini APIs for custom AI assistant functionality, replacing ChatGPT Enterprise’s custom chatbots. HR teams built compliance-specific tools in isolated Lovable and Supabase environments to meet regulatory requirements. The international team created a backend service that normalizes property data from regional CRMs, a complex integration that previously would have required months of vendor development.

Rollout Timeline and Organizational Adoption

eXp discovered Lovable in early 2025 and moved to full organizational adoption within months. The Peru website incident accelerated the international infrastructure migration, which was completed across 26 countries by late 2025. “The Hub” launched at eXp’s annual Miami conference, replacing Slack’s community functionality for the full 83,000-agent organization. Internal support ticket volume dropped over 70% as teams migrated workflows into Lovable-built tools.

The company now operates dozens of custom applications, many serving single teams with requirements too specific to justify traditional development resources. This long-tail software strategy building tools that never would have reached an engineering roadmap represents a second-order effect of democratized development capabilities.

Also Read Intel MWC 2026: Live AI Inference Redefines Mobile Network Architecture

Frequently Asked Questions (FAQs)