Many technology companies are clamoring to claim an artificial intelligence (AI) basis for their product, especially given the pull AI products seem to have with investors , but many fail to consider whether AI is actually the best solution to the problem they're hoping to solve. When TurboPatent announced that its latest patent drafting software, RoboReview, would be an AI-enabled product, I initially assumed that the product would fall into this bucket of startups frantically trying to weave the words "artificial intelligence" into its marketing.

However, patenting may actually be one of the best fits for artificial intelligence in the legal profession patents are rife with technical language and often are held up at the U.S. Patent and Trademark Office (USPTO) by basic procedural writing gaffes, which makes them a great candidate for algorithmic sorting.

Here's a look at why RoboReview may be a somewhat rare case of AI technology being best suited, rather than just tossed at, a specific legal problem:

Who it serves: The tool is designed for patent drafters, but TurboPatent founder and CEO James Billmaier explained that the tool was created with an eye to reducing costs for smaller startups and businesses that may not have the resources to pay an expensive patent attorney. Those businesses, he noted, shouldn't forgo an attorney altogether and try their own hand at drafting a patent. Instead, he suggested handing this product to their attorney.

What it does: RoboReview uses four "expert bots" that read through a patent application prepared in Microsoft Word and flags any problems with antecedent basis, figures references, claim support, and claim order and formatting. Using almost 20 years of the USPTO's published patent applications, these bots are trained to identify both basic word ordering errors and larger conceptual failings, things like inconsistent or incomplete definitions for key terms.

After these "expert bots" get through with your document, they return it in Word with a set of tracked changes and comments, much like if you'd gotten feedback from a human reader. "The results look exactly like if you sent your document around, and a bunch of smart people made comments except in this case it's just a bunch of robots doing that but with much more accuracy and speed," Billmaier said.

Does it work? The question with any AI product is whether it identifies the right things, rather than just a random set of things that a human is then charged with correcting by hand. I'm not a patent attorney, so I don't know that I can answer that question definitively, but from what I saw, the tool can use the context of your document to identify problems beyond simple grammar and language errors you get from many tools on the market. "Previous attempts to do this have just been word matchers, so you miss the conceptual inaccuracies," Billmaier explained.