Real Estate

AI incident summaries for a property security team

AI
Laptop showing a sample security incident report. On the left, an officer's narrative has names, locations, a badge number, and a license plate marked for removal. On the right, an AI summary describes the incident in four short sentences, with tags confirming the personal details were removed and the weapon reference was kept.

About Retail Property Management Company

Our client operates a large portfolio of retail properties across the United States, protected by a security team of about 2,000 dispatchers and officers. We maintain and extend the operations platform their properties run on, including the tools the security team uses to log and review incidents.

Incident reports that take too long to read

Long narratives, sensitive details

Officers write incident narratives in their own words, and a single report can run long. Dispatchers and managers need the key facts fast. Every report also carries names, locations, and other identifying details that shouldn't travel further than they need to.

‍

A feature inside a platform we already know

We already maintain the platform where these reports live, so we knew the data, the users, and the release process. That let us scope an AI feature that fit the existing workflow, with no new tool for officers to learn.

Summaries on Azure OpenAI, with personal details removed

From narrative to synopsis

We built incident summarization on Azure OpenAI using GPT-4o-mini, a fast, cost-efficient model suited to short text. It takes an officer's narrative and returns a concise synopsis with names, locations, and other identifying details stripped out.

‍

Tuned for security work

Security reports describe violence and weapons, and Azure OpenAI's default content filters can reject that kind of text. We applied a custom content filter at the most permissive thresholds, so genuine reports get summarized instead of blocked.

Live for 2,000 dispatchers and officers

In production

The feature launched in production in September 2026 and is available to the client's full security team of about 2,000 dispatchers and officers. Early reports are already running through it.

‍

What's next

Today the feature works on typed narratives. Document processing is next on the roadmap, so attached reports and forms can be summarized the same way.

Ready to get started?

We see our clients as our partners and are invested in their success. We ensure what we build for our clients will have a positive ROI. We never build anything just for the sake of it.