SafetyAI: State of the Art AI for Construction Safety
Since 2017, we conduct cutting-edge research and development to bring the latest AI innovations to construction safety. We are 100% supported by the SafetyAI Council, a consortium of leading companies that fund our work, share their datasets for collaborative machine learning, and test our models and applications in the field.
ChatSafetyAI
ChatSafetyAI is a custom ChatGPT tailored for construction safety, designed to assist professionals with all their OHSE-related tasks. ChatSafetyAI accepts text queries in English, Spanish, and French, as well as photo and file uploads. It is also available as an API.
ChatSafetyAI is under active development since February 2023!
See recent updates:
Q1'24,
Q2'24,
Q3'24,
Q4'24,
Q1'25.
👉 Request access to the demo here.
HECA from Photos
Injury Prediction
Safety Paperwork Filling
Presentation Generation
Climate Survey Analysis
OHSE Regulations Query
ChatSafetyAI on Smartphones
Vision and Scientific Expertise
ChatSafetyAI is developed by leading researchers in AI and construction safety, with collectively over 200 peer-reviewed papers and 14k citations since 2005. Long before the AI boom, Dr. Antoine Tixier and Dr. Matt Hallowell led the way in applying machine learning to construction safety data.
Made by the Industry
The SafetyAI Council, a team of major companies, provides use cases to customize modes like Regulations, Photo HECA, and Pre-job Brief. They also share incident datasets to train Super Predictive Models, boosting forecasts of risk and injury characteristics (Safer Together initiative).
Built for the Workplace
ChatSafetyAI features guardrails that block non-work-related queries, keeping usage focused and professional. It automatically blurs faces, redacts sensitive text, and never uses conversations for training. It can be deployed in-house for maximum privacy and security (on Azure or AWS).
Flexible & Expandable
Modes can be customized per company and accessed as APIs. ChatSafetyAI can have knowledge of your documentation, generate documents and presentations using your templates, and integrate with your systems.
Engineered Knowledge & Prompts
The SafetyAI R&D team curates cutting-edge construction safety knowledge and crafts precise prompts. This ensures ChatSafetyAI understands domain-specific jargon and provides reliable answers.
Best of Each LLM
ChatSafetyAI capitalizes on the respective strengths of multiple cutting-edge LLMs. This ensures optimized performance in image understanding, object detection, long context handling, and other specialized tasks.
SafetyAI Dashboard
The SafetyAI Dashboard is an interactive web application for analyzing and visualizing a company's incident and observation dataset (narratives + safety outcomes) using natural language processing, network analysis, time series analysis, and other data mining methods. The Dashboard also provides an interface to the SafetyAI predictive models (company custom and community), and to a binary and semantic narrative search engine. All the methods the Dashboard relies on were developed as part of multiyear research projects and have led to peer-reviewed publications in AI and construction journals and conferences. All methods have been developed and validated on large-scale, real-world incident datasets.
Stats & Time Series Forecasting
Word Networks
Attribute Networks
Risk Assessment
Predictive Models
Binary & Semantic Search
Learning Injury Precursors
SafetyAI Assistant
The SafetyAI assistant is a simplified version of the SafetyAI Dashboard that features the functionalities that are the most useful onsite, with a user interface optimized for smartphones, and voice recognition capabilities. By answering a few simple questions about work conditions and the task at hand for a particular work period, the SafetyAI Assistant provides on-site personnel with: (i) a risk assessment, (ii) predictions of the most likely accident characteristics, should an incident occur, and (iii) historical injury cases that have occurred in similar work in the past. This information is very useful during, e.g., pre-job safety meetings.
SafetyAI Assistant Demo
Team
Dr. Antoine Tixier
Co-Founder and CEO/CTO
Antoine's PhD (2013-15) focused on how to predict the characteristics of construction accidents with machine learning.
Between 2015 and 2020, while a postdoc researcher in deep learning and natural language processing, Antoine started to work on SafetyAI in his free time. He built on his PhD research to develop new incident data analysis and visualization techniques and better prediction methods, and created the SafetyAI Dashboard app.
In 2020, Antoine started working full-time on SafetyAI and created the SafetyAI Assistant app and the Safer Together approach (community predictive models). At the beginning of 2023, Antoine created ChatSafetyAI, the ChatGPT for construction safety.
Antoine is in charge of every aspect of SafetyAI, from conducting research, training models and writing papers, to developing web apps, creating online content (tutorial videos, website), and interfacing with clients (demos, support, reporting, in-house deployment, admin and legal aspects)...
Dr. Matthew Hallowell
Co-Founder and Strategic Advisor
Matt is a President’s Teaching Scholar and Endowed Professor of Construction Engineering at the University of Colorado at Boulder. He earned a BS and MS in Civil Engineering and a PhD with a focus on Construction Engineering and Occupational Safety and Health. Before his academic career, he worked in construction as a laborer, project engineer, and quality inspector.
Matt specializes in construction safety research, with an emphasis on the science of safety. He has published over 100 peer-reviewed journal articles on energy-based hazard recognition, safety leading indicators, safety risk assessment, predictive analytics, and precursor analysis. For his research, he has received the National Science Foundation CAREER Award, the inaugural Thomas F. Farrell, II Safety Leadership and Innovation Award, and the Construction Industry Institute Outstanding Researcher Award.
SafetyAI Council
Joining the SafetyAI council is the opportunity to participate in an industry R&D consortium with collaborative benefits: member companies are an integral part of our research and development process, by providing use cases, sharing accident datasets to improve injury prediction models (Safer Together approach), influencing research directions, and testing web apps in the field.
Updates from the most recent council meetings:
Q1'24,
Q2'24,
Q3'24,
Q4'24,
Q1'25.
Safer Together Initiative: after anonymization and normalization, predictive models are trained on the aggregated datasets of the Council members. These community models (aka "Super Models") outperform the company-specific models at predicting injury characteristics most of the time, while also generating finer-grained forecasts. The SafetyAI Council's Super Models tackle the holy grail of safety cross-organizational learning and dissemination in the construction industry. Models are available in all of SafetyAI's web apps, and as an API.
Pricing Tiers
SafetyAI is entirely member-funded, with no external investors, to keep our research focused on long-term benefits, rather than on short-term profit. For this model to work, each organization needs to contribute proportionally to its capacity to support and benefit from innovation.
$6k /year
Small
For small companies to use ChatSafetyAI without sitting on the Council.
- 1B tokens of ChatSafetyAI
- Updates and support
- Community predictive models
$11k /year
Medium
For medium companies to use ChatSafetyAI without sitting on the Council.
- 2B tokens of ChatSafetyAI
- All of the small tier, plus:
- In-house deployment
$19k /year
Large
For large companies interested in supporting and participating in R&D.
- 4B tokens of ChatSafetyAI
- All of the medium tier, plus:
- Seat on the SafetyAI Council
- SafetyAI Dashboard & Assistant
- Custom predictive models
$32k /year
Big
For big companies interested in supporting and participating in R&D.
- 8B tokens of ChatSafetyAI
- All of the large tier, plus:
- Push your use cases in priority
- Highest level of customization in ChatSafetyAI
Built by Antoine Tixier based on this template.