Sinyi Realty’s agents struggled to present cluttered properties and create professional marketing materials efficiently, impacting sales opportunities across Asia.
Sinyi deployed FLUX.1 Kontext [pro] on Microsoft Foundry to remove clutter from property photos and launched AI FlyerGEN to automate branded flyer creation.
Agents now attract more sellers and save time with automated marketing workflows. Sinyi plans to expand AI solutions to all listings and future sales materials.
Every real estate agent faces the same dilemmas: How do you showcase a property’s potential if the current owner’s clutter fills every room? And how do you create professional marketing materials without sacrificing time that could be spent with clients? For Sinyi Realty’s thousands of agents across Taiwan, China, Japan and Malaysia, these challenges were adding up to lost opportunities.
Founded in Taiwan, Sinyi Realty has evolved from a traditional real estate brokerage into a technology-powered service company determined to solve these problems. With approximately 3,500 agents serving customers across multiple markets, the company has embarked on a comprehensive digital transformation, leveraging AI, big data and cloud infrastructure to enhance both customer experience and internal operations.
“We are not just a traditional real estate brokerage but a technology-empowered real estate service company,” explains ChuTing Hsu, Project Manager on Sinyi’s AI team. “Our guiding philosophy is that AI doesn’t replace humans—it empowers them. Technology is treated as a tool to improve transparency, efficiency and trust.”
“Our guiding philosophy is that AI doesn’t replace humans—it empowers them. Technology is treated as a tool to improve transparency, efficiency and trust.”
ChuTing Hsu, Project Manager, Sinyi Realty
Sinyi implemented a custom-built “Clear All” feature on the company’s real estate platform, powered by Black Forest Labs’ FLUX.1 Kontext [pro] model on Microsoft Foundry. The workflow begins with an Azure OpenAI model on Microsoft Foundry that analyzes interior property photos to determine whether an image depicts an indoor space and then identifies noticeable clutter. If the AI detects clutter, the photo is processed using Kontext [pro], which removes unwanted items.
Sinyi discovered Kontext [pro] and, recognizing its potential, worked with Microsoft to deploy it on Foundry. Sinyi was already using Azure OpenAI models and wanted to maintain a secure, unified workspace for data operations. Foundry's security capabilities allow the company to protect sensitive customer data while streamlining AI workflows.
“We wanted to stay in the same workspace rather than having to maintain different platforms,” explains TeWei Hsieh, who works in data engineering and data architecture. “By keeping FLUX Kontext model in Foundry, our data scientists and data engineers can work in the same environment.”
The team chose the model specifically for its approach to image transformation. Unlike models that completely regenerate images, Kontext [pro] works from the existing photo, selectively removing clutter while preserving the original structure.
“FLUX Kontext has better performance for cleaning spaces in our scenario,” says TeWei Hsieh. “It doesn’t generate a similar space without clutter; instead, it works from the original photo, making users feel that unnecessary items are simply removed from it.”
For flyer creation, Sinyi developed AI FlyerGEN, which leverages Kontext [pro] models from Foundry to automate the design process. Agents can create two types of flyers: personal marketing materials to promote themselves or property-specific flyers to showcase homes for sale. The system includes built-in brand templates and style guidelines including fonts, colors and logos to ensure consistent brand identity. The AI automatically adjusts copy based on property information and can generate professional profile photos for agents. The system uses Azure Databricks to run Python programs that process PowerPoint templates and generate images and branding photos. Agents simply select the flyer type and upload their data to generate fully designed flyers in seconds.
At the heart of Sinyi’s AI workflow is Azure Data Lake Storage, which serves as the single source of truth for all AI-generated and processed assets. The company manages approximately 1 petabyte of data across all operations, with about 10 gigabytes dedicated to images for their AI projects. The storage solution integrates with Azure Blob Storage, and by mounting both as volumes within Azure Databricks, data scientists can work seamlessly within a unified environment without switching between tools.
The storage holds flyer templates, ETL outputs, AI-generated agent photos, structured data and metadata used for flyer composition. That metadata includes salesperson awards, relevant certifications, property information and geography descriptions. By integrating Azure Data Lake Storage with Azure Databricks and mounting it as a volume, data scientists can read from and write to storage directly within Azure Databricks notebooks. This seamless integration eliminates the need to switch between different tools and platforms, ensuring smooth workflow efficiency throughout the development and deployment process.
Sinyi has also embraced Microsoft Fabric with Power BI as its business intelligence foundation, managing a portfolio of over 350 Power BI dashboards that support both sales and customer analytics. The company oversees the complete data flow from initial ingestion through complex transformation—utilizing Data Factory pipelines for orchestration and Spark for high-scale processing—to the final data modeling for visualization.
A key innovation in Sinyi's approach has been implementing Power BI Deployment Pipelines using Fabric Capacity. This implementation empowers the company's non-IT business partners to execute agile, self-service report iteration while enforcing strict version control to preserve institutional business knowledge. The deployment pipeline structure allows reports to move systematically through development, testing, and production stages, ensuring quality control while enabling business users to work more independently.
The team also uses its data platform for proactive data security monitoring, ensuring the integrity and compliance of their analytical assets across the organization.
The Clear All feature is now live and rolled out to select listings. Sales agents have reported that the feature helps them attract sellers to sign contracts by presenting properties in their best light. Sinyi plans to expand the service to all listings and will monitor its impact on inquiry volume and website dwell time.
The Minimum Viable Product is expected to be piloted across business regions by the end of October 2025, and the full launch for all sales representatives is scheduled for 2026.
“Creating customized flyers used to be time-consuming, but with this workflow, the entire process can be nearly fully automated,” says ChuTing Hsu. “All agents need to do is upload data and click a few buttons. This allows them to focus on what truly matters: building relationships and closing deals.”
Sinyi continues exploring AI applications, including potentially expanding automation to sales reports and property introduction materials.
“As AI technology becomes more mature and its applications more widespread, many tasks that previously required manual handling will gradually be replaced by AI in the future,” ChuTing Hsu says.
Discover more about Sinyi Realty on Facebook.
Follow Microsoft