It’s easy to dismiss artificial intelligence and machine learning as just buzzwords, or technology that belongs in the distant future.
But that’s not the case. Here at The Red Fox Group we’re utilising advanced machine learning and artificial intelligence techniques to drive efficiencies for our customers, even in places where you might least expect it.
We’ve recently deployed a new version of the business card scanner in SwiftFox CRM which now utilises some very exciting cutting-edge features.
Optical Character Recognition is a technology that has been around for decades. Computers have long been able to read text and convert it to digitised information. We’ve taken that to a new level by partnering with Microsoft to utilise their Computer Vision cognitive service.
Computer Vision is OCR technology on steroids. It’s enhanced with machine learning models so SwiftFox can now make intelligent assumptions about the nature of objects and characters rather than just trying to map it using simple pattern recognition.
This gives it huge advantages. It can handle blurred text, extravagant fonts, and text at arbitrary angles – even upside down. It's much more accurate because of that. If you’re scanning a business card in real time at a cocktail function, or in the Uber on the way home, this will radically improve the accuracy of text recognition.
But that’s just the start.
Once we have captured the text out of an image of a business card we’re able to use machine learning text analysis to figure out how the data is structured and how it relates to our data models for people and organisations.
We get both Microsoft and Google’s cloud AI services (Microsoft Text Analytics and Google Cloud Natural Language) to analyse the grouped text structures and pass back which data fields each piece of text should map to, and how confident the artificial intelligence is of its finding.
SwiftFox then analyses the results of the two-machine learning models and uses the data that is most accurate to pre-populate data fields.
A very common problem with traditional business card scanning technology is that it has made a lot of assumptions about the text layout – eg most business cards have the person’s title directly after their name. But what if the business name comes after the person’s name? Or their job title comes first? We’re using machine learning to bypass assumptions and are getting far more accurate data back on what information is captured in each piece of text.
This means that when you’re loading a new contact through the business scanner, the chances are really good that we’ve correctly identified which parts of the business card go into the right part of the database. No more accidentally creating new organisations called ‘Managing Director’!
Hopefully this gives you a better idea of how we’re using advanced artificial intelligence and machine learning to make even simple tasks more efficient.