For the engineering team at Multitude Insights, there’s no such thing as a “quiet day.” Whether it’s planning the next BLTN feature, or coding for the imminent UI update, it’s a fast-paced environment.
We know that each hour we dedicate to perfecting our software might save hundreds of hours for frontline police officers fighting to keep the US public safe. The latest feature we’re bringing to BLTN, Smart Link, might deliver the greatest efficiency gains yet.
Smart link is an AI-driven analysis tool that examines information on the BLTN platform, finds connections, and generates detailed insights for investigators. By automating this analytical work, Smart Link will cut down time spent on office-based work, boost detection rates, and better reward accurate information-sharing among police departments.
The thinking behind Smart Link
We are hyper-focused on making it easier for police officers to be more productive and collaborative. The best way for BLTN to achieve this is to flag opportunities for collaboration and find connections among all the data on the platform. That's where PoiRoBot came from: that’s the name we gave our AI investigative assistant (a tribute to Agatha Christie's famous fictional detective).
The categories that we focus on when it comes to matching reports are: "location," "suspect," "vehicle," and "image.” These are the most common dimensions that provide a link between incidents, and categorizing the connections found in this way helps our users understand how PoiRoBot made a decision to link two reports. This is part of what we call "explainable AI": it’s our method to remove the black box from AI and provide the rationale behind the autonomous matching system to our customers. Another dimension to this is providing multiple ways for our users to view these connections. The map view shows geographic distance, the tabular view allows users to cross-reference the different dimensions, and the evidence board shows the strength of the connections - the closer the objects, the stronger the connection.
The technical challenges
Technically, we asked our AI specialists to bring this vision to life, and although the solution was complex, our goal was to make it feel simple to the user. We use multiple machine learning models in tandem, as well as a CJIS compliant large language model (LLM) to make this work. It truly is amazing technology, and we are very proud of our stellar team for putting it together. Another exciting feature of our technical process is that it gets better over time: when users interact (or don't) with one of PoiRoBot's links, we track that and use that to make PoiRoBot better.
The Future:
This technology is just going to get better and better. The more experience PoiRoBot has, the better it will be, and the more value it will deliver to BLTN users. Its incredible scalability means that it can even take on large problems at a national or global scale. The sky's the limit, really, and we can’t wait to see it evolve. Our hope is that with this technology in hand, officers and crime analysts will be able to catch more criminals, and that both their communities and their jobs will be safer as a result.
Over the past decade or so, technological progress in society has often felt more like a hindrance for law enforcement than a tool for forging a safer society. With each new major development, whether it’s smartphone, social media or cyber related, has come a ton of extra day-to-day work for investigators to cope with. Even the useful new hardware, like body cameras, has come with an additional administrative time cost. It’s time for law enforcement’s technology partners to address this mounting problem, and, for Multitude Insights, BLTN’s new Smart Link feature is an important step forward.
.jpg)
