What is Safety AI?

Updated December 30, 2025 23:04

What is Safety AI?
How do I enable Safety AI for my company?
How Does It Work?
How will I find out if there is a risk identified in my walk or pano?
How can I interact with risks detected by Safety AI?
Who receives the email notifications?
Who can see safety risks in DroneDeploy?
What is the turnaround time for a Safety AI report?
What types of safety risks can Safety AI detect?

What is Safety AI?

Safety AI is an intelligence add-on for DroneDeploy Ground. It scans uploaded 360 walkthroughs and panoramas to detect hazards. Each identified risk is saved as a Field Note, allowing teams to resolve issues privately before they lead to incidents.

Safety AI requires no additional setup or training. There is no limit to the number of walks or panos you can upload, as every walk and pano you capture is processed using Safety AI. We can also run Safety AI retroactively on walks or panos that were performed in the past.

How do I enable Safety AI for my company?

Safety AI is only available to Enterprise customers as an add-on to DroneDeploy Ground. Please contact your Account Executive or Customer Success Manager to set up a trial! You will get the most value from your trial by selecting projects that capture frequently and consistently in DroneDeploy Ground.

After Safety AI has been enabled for your organization, you will be able to activate Safety AI on any project using the following steps:

Open the DroneDeploy web app and navigate to your project.
Click the gear icon in the top right corner to open the project settings.
A toggle in the “Safety AI” section will be turned “off” by default.
Switching this toggle “on” prompts new walks and panos captured on this project to be automatically processed by Safety AI.
1. Note: Organization administrators and project coordinators can manage which users have Safety AI access on the project through the feature provisioning modal.
Switching this toggle “off” means that Safety AI will not process new walks and panos uploaded to this project, and the feature provisioning modal will no longer be accessible. Field notes that Safety AI previously created will still be visible to users who were given Safety AI access on the project.

Please note that you must be an organization administrator and/or project coordinator to enable Safety AI at the project level.

Organization administrators can enable Safety AI on any project in their organization.
Project coordinators can enable or disable Safety AI on any project where they have coordinator permissions.

How will I find out if there is a risk identified in my walk or pano?

If there are safety risks identified in your walk or pano, those risks are summarized into a PDF report that is emailed to your team by 7:00am (local time) the following morning. The PDF report will include the following details for each risk:

A cropped image of the risk from your walk.
An AI-generated description that includes a reference to OSHA Standard 1910 or 1926.
A map to show where the risk was identified on the floor plan
A link to the Field Note in the DroneDeploy web app.

How can I interact with risks detected by Safety AI?

Each risk is saved as a Field Note in DroneDeploy, which allows you to add comments about the Field Note, close the Field Note once resolved, or delete the Field Note if it does not apply to your company’s safety program.

Who receives the email notifications?

Reports are emailed to your Safety Team, which is composed of DroneDeploy users who have been added to your project in DroneDeploy and enabled for Safety AI access through the feature provisioning modal.

Who can see safety risks in DroneDeploy?

Visibility of safety risks is controlled by the same feature provisioning modal as the email notifications.

What is the turnaround time for a Safety AI report?

Turnaround times vary based on the volume of media uploaded. The average turnaround time from upload to delivery is approximately 14 hours for walks and 7 hours for panos.

What types of safety risks can Safety AI detect?

Our models are trained to detect and describe the most common safety risks found on construction sites. Risks are sorted into 10 categories of general safety risks and there are several subcategories of safety risks that we are able to identify. For example, in the “Personal Protective Equipment” category, we verify that each person captured in a walk or pano has a hard hat, safety glasses, high visibility clothing, safety gloves, closed-toe shoes, and, if applicable, face protection.

To customize Safety References and Risk Filtering, please refer to Customization for Safety AI

Included below are examples from the remaining nine categories of safety risks that we can detect: