Revolutionizing Construction with AI: Lessons from Toga AI

Table of Contents

1. Introduction
2. The Challenge in the Construction Industry
3. Building an Organization to Solve the Challenge
4. Technical Aspects of the Solution
   1. Segmentation Model
   2. Object Detection Model
   3. Room Classification Model
   4. Understanding the Structure of the Building
5. Lessons Learned
   1. Start Where They Are
   2. Build Custom Code Only When Necessary
   3. Machine Learning is Far from Automated
   4. AI is Not Possible Without IT
6. Conclusion
7. FAQ

AI Meets Real World: Deploying an AI Solution in the Construction Industry

The construction industry is one of the most important industries in the world, responsible for building the infrastructure that we all rely on. However, it is also an industry that has been slow to adopt new technologies, and is often resistant to change. In this article, we will explore how we deployed an AI solution in the construction industry, and the challenges that we faced along the way.

Introduction

Alexander Parowska, the CTO of Toga AI, has worked as a software and data architect for more than 15 years. Toga AI is a startup that is bringing AI to the construction industry, with a focus on pre-construction space. In this article, we will explore how we built an organization to solve the challenges in the construction industry, and the technical aspects of the solution that we developed.

The Challenge in the Construction Industry

The construction industry is a complex and challenging industry, with many inefficiencies in the pre-construction phase. Architects typically design buildings, and then send out PDF plans to general contractors and subcontractors. This process is time-consuming and inefficient, as each party needs to estimate the cost of their part of the project. Toga AI identified this as a challenge, and set out to build a solution that would automate the process.

Building an Organization to Solve the Challenge

Toga AI started by building a team of machine learning experts through Tribe AI, a network of top global technical talent. The team focused on building an automated takeoff software solution that would be web-Based. They built a marketing Website using a no-code tool, and focused on building custom code for the application. The team relied on machine learning models to solve the challenges in the construction industry.

Technical Aspects of the Solution

The team at Toga AI built several machine learning models to solve the challenges in the construction industry. They built a segmentation model to discern which pixels were walls and which were not. They also built an object detection model to extract semantically Meaningful data from the floor plan. Additionally, they built a room classification model to detect different types of rooms, such as bedrooms, bathrooms, and closets. Finally, they built a model to understand the structure of the building using graph neural networks.

Lessons Learned

The team at Toga AI learned several lessons along the way. First, they learned that it is important to start where the industry is, and then try to lure them in from there. Second, they learned that it is important to build custom code only when necessary. Third, they learned that machine learning is far from automated, and requires deep expertise in computer vision and machine learning. Finally, they learned that AI is not possible without IT, and that software development is a Core part of building an AI solution.

Conclusion

In conclusion, deploying an AI solution in the construction industry is a complex and challenging task. However, with the right team and the right approach, it is possible to build a solution that can automate many of the inefficiencies in the pre-construction phase. The team at Toga AI learned many lessons along the way, and their experience can be applied to other industries that are interested in deploying AI solutions.

FAQ

Q: What is Toga AI? A: Toga AI is a startup that is bringing AI to the construction industry, with a focus on pre-construction space.

Q: What challenges did Toga AI face in the construction industry? A: Toga AI faced many challenges in the construction industry, including inefficiencies in the pre-construction phase and resistance to change.

Q: What machine learning models did Toga AI build? A: Toga AI built several machine learning models, including a segmentation model, an object detection model, a room classification model, and a model to understand the structure of the building using graph neural networks.

Q: What lessons did Toga AI learn? A: Toga AI learned several lessons, including the importance of starting where the industry is, building custom code only when necessary, and the fact that machine learning is far from automated.

Q: What is the future of AI in the construction industry? A: The future of AI in the construction industry is bright, with many opportunities to automate inefficiencies in the pre-construction phase. However, it will require deep expertise in computer vision and machine learning to build effective solutions.