Security teams often struggle to track unused user accounts across multiple systems, creating compliance risks. Operator can automatically audit user access logs, flag inactive accounts, and even disable them—without IT intervention.

Best Practices for Automated Compliance Audits

• Set clear access policies. Define inactivity thresholds (e.g., 90 days of no login) in your prompt. Don’t forget, Operator can do math, so the thresholds can be calculated.
• Use AI to analyze patterns. Operator can spot anomalies in access logs.
• Automate remediation actions. AI can generate alerts or trigger account deactivation.
• Document audit trails. Maintain logs of changes for compliance reporting, this can be done in Google Sheets or other web-based applications. 

Example Prompt:

Visit the page "{your dashboard here}", sort the users by "Last LoggedIn", and list the usernames and emails of every user who have never logged in and every user that hasn't logged in within the last three months.


Routine preventative maintenance is crucial for ensuring optimal building performance, but manually reviewing schedules and equipment runtimes can be time-consuming. With OpenAI Operator, you can automate preventative maintenance checks in Niagara by reviewing HVAC schedules and identifying inefficiencies before they become costly issues.

Best Practices for Automated Maintenance Checks

• Use full URLs to work around navigation challenges. Niagara’s navigation is difficult for AI to navigate, so you may want to just give a list of URLs to the application to simplify the process. 
• Be very prescriptive. The Operator can easily make mistakes without explicit data to review, so have it perform actions like clicking on schedule items to check start and end times numerically instead of just reviewing the visuals.
• Generate reports for maintenance teams. Automate notifications and logs for review by having Operator open a web-based email application and sending a note.

Example Prompt for HVAC Preventative Maintenance


Visit '{Neeve remote access portal url}', wait for it to load. Then visit '{Neeve remote access full path to Niagara AHU 1 schedule page}'

Carefully review the column for each day of the week. Click the mouse in the column for each weekend day (Saturday and Sunday). When you click the column the Event Start and Event Finish fields should be greyed out and have 12:00AM as values for both fields. If not, that is a violation. Check each week day (Monday - Friday). When you click the column the Event Start and Event Finish fields should be active (not greyed out) and the value for Event Start should be no earlier than 6:00AM and the value for Event End should be no later than 8:00PM - otherwise it is a violation.

Repeat that for these URLs:
'{Neeve remote access full path to Niagara AHU 2 schedule page}'
'{Neeve remote access full path to Niagara AHU 3 schedule page}'
etc.

Summarize the schedules for each AHU and tell me which ones are in violation and why.

Many enterprise systems lack APIs, making it difficult to extract data for reporting and analytics. OpenAI Operator can interact with web-based or desktop applications just like a human, automating data scraping and report generation with natural language commands.

Best Practices for Automated Data Extraction

• Use clear and structured prompts. AI works best with well-defined instructions.
• Break down workflows into steps. Identify login sequences, navigation paths, and the required data fields.
• Leverage AI for formatting. Operator can clean, summarize, and format extracted data automatically.
• Use Google or Office for data storage: Operator cannot download files and is only able to access things through the web, so leverage Google sheets or similar programs for data storage. 

Example Prompt for Automated Reporting


Visit "{Your Neeve portal url}", wait for it to load. Then visit the URL "{Neeve remote access device link}" and log in to the APS-90E device using the credentials. Access the "Start Page - Live View", obtain the "In" number and "Out" number for adults, subtract the "In" from the "Out" number, and enter today's date and the result into a Google Sheet at "{Google Sheet URL}", recording the date in the "Date" column and the result in the "Occupancy" column. Do not ask confirmation as to whether you should update the sheet, just proceed. Close the tab that has the page for the sensor when you are complete.


Building automation systems like Niagara control thousands of data points across HVAC and environmental systems. While FDD systems are available in the market, they may not be cost effective for smaller installation. OpenAI Operator can provide basic fault detection and diagnostics (FDD) by automatically comparing HVAC setpoints to actual room temperatures, flagging potential performance issues.

Best Practices for AI-Driven Fault Detection

• Compare expected vs. actual performance. Identify zones where setpoints are not being met.
• Spot early warning signs. Catch underperforming HVAC zones before complaints arise.
• Generate reports for facility managers. Automate alerts for prioritized maintenance.

Example Prompt for FDD Analysis


Visit '{Your Neeve portal url}', wait for it to load. Then visit '{Neeve remote acccess full path to VAV Summary}'



On this page, for each VAV in the column, check if the Space Temp is higher than the Space Setpoint. If so, this could be an issue and remember the name of the VAV. In the end, give me a list of all the VAVs that are potential issues and why.