Recently featured in ENR, Intraspexion is a deep learning startup whose CEO Nick Brestoff is a former engineer (but spent the last 38 years as a litigator!) Back in January, the US Patent & Trademark Office issued it a patent for “using classified text and deep learning algorithms to identify risk and provide early warning.” Although the company’s seven patents are primarily aimed at identifying discrimination and products liability matters before they turn into full-blown claims or litigation, there could be a construction application as the company grows and scales.
Currently, Intraspexion is piloting its software with corporate legal departments who recently settled an employment discrimination lawsuit which lasted long enough for the company to produce emails in Outlook format to the plaintiff, and which included “risky or even ‘smoking gun’ emails.” Intraspexion claims that it will identify those emails (among others, potentially) and, on its website, asks interested companies whether they would have had more time to investigate and mitigate the risk and expense of litigation had they applied Intraspexion’s technology.
According to ENR, Intraspexion is still pre-revenue but has a “functional system” and is working on a pilot project with a public transportation conglomerate (whose name Brestoff did not provide to ENR). However, he did tell ENR that “if you know of a construction company that’s looking at Deep Learning now, I’d love to work with one.”
“Deep Learning is a multi-layered neural network that also learns from mistakes and back-propagates to minimize errors, so, perhaps in the construction industry, there may be many text-based examples of warnings in post-incident reports,” Brestoff told ENR. “There may be a thousand post-incident reports concerning some similar category. Could any engineer working on similar project remember them? No. But a Deep Learning model of that risk can ‘pattern match’ for it. It augments your intelligence so you can go from managing a lawsuit to investigating the risk of one, and go from being reactive, to being proactive.”
Machine learning and artificial intelligence aren’t exactly the same thing, although the terms seem to be used somewhat interchangeably as you read more about the topic. But as a rule of thumb, AI is a broader concept under which machines have the ability to carry out tasks or functions like a human in a way that we would consider, well, intelligent (like a “smart” stock picking function or driverless cars). Machine learning, on the other hand, is a narrower application, where we give machines data and they have some ability to learn from, and then apply, that data themselves.
Obviously the applications for a deep learning model applied to construction contracts, insurance policies, project risk management and claims, and other AEC industry applications are myriad, and could change the way the industry thinks about how to manage risk. For one other current example of an applied deep learning algorithm, check out my friend Chris Cheatham’s company RiskGenius, which is doing some amazing things in the insurance space.
Happy Thanksgiving from AEC Labs!
