{"id":3498,"date":"2023-10-09T14:28:52","date_gmt":"2023-10-09T14:28:52","guid":{"rendered":"https:\/\/nilg.ai\/?p=3498"},"modified":"2023-10-16T13:34:17","modified_gmt":"2023-10-16T13:34:17","slug":"nilg-ai-success-hodcon-2023","status":"publish","type":"post","link":"https:\/\/nilg.ai\/pt\/202310\/nilg-ai-success-hodcon-2023\/","title":{"rendered":"NILG.AI’s Success Story with Vonovia at the Hands on Data Conference 2023"},"content":{"rendered":"

\"Bastian<\/p>\n

In the dynamic world of AI, NILG.AI continues to shine as a beacon of innovation and excellence. At the<\/span> Hands on Data Conference 2023 (HODcon)<\/span><\/a>, our spotlight shone on our journey on the \u201cImage Insight\u201d Challenge with<\/span> Vonovia<\/span><\/a>, Europe’s leading housing company. Our CEO, Kelwin Fernandes, took the stage alongside<\/span> Bastian Winkenbach<\/span><\/a>, Head of Advanced Analytics at Vonovia, to share NILG.AI’s Success Story on this challenge. But guess what? The path was not so straightforward.<\/span><\/p>\n

 <\/p>\n

So here comes the story of the \u201cImage Insight\u201d Challenge<\/span><\/h2>\n

The purpose of the \u201cImage Insight\u201d Challenge was to recognize damages in tenants’ reports to minimize delays and enhance customer support, but <\/span>no data was available<\/b>. So, when we were building the proposal, we constructed a dataset from images scraped from the internet to ensure that if we were chosen, we had the tools needed to carry on with the project successfully.<\/span><\/p>\n

And we won, Yay!\u00a0<\/b><\/p>\n

The project had already started, and there was a sudden setback: Vonovia’s legal team alerted us of their preference for <\/span>not using public data<\/b> scraped from the net. It was best to play safe. <\/span>What could we do? The project was already going!<\/b><\/p>\n

Our solution to a major data issue<\/span><\/h2>\n

At the time, we were (and still are) right at the center of the generative AI hype, and we so right into this opportunity:<\/span> Using generative AI to generate synthetic data<\/b><\/a> of appliance pictures and emails from tenants complaining about issues in their flats. And just like that, we managed to train the initial models with synthetic data while<\/span> preserving the tenants’ privacy<\/b><\/a> and ensuring the project’s success.<\/span><\/p>\n

Want to know more about our solution with generative AI? Check our videos:<\/span><\/i><\/p>\n