Natural language processing is coming to a logistics application near you
Here’s a thought exercise: think, really think, about a typical day in your office.
Think about how much time you spend typing on a computer. What are you typing? Are you opening emails? Searching the internet for information? Copying and pasting? Pulling data from one system into another?
Now think about the speaking part of your day. Talking to co-workers, phone calls, teleconferences, meetings.
Despite all that’s been written about the need to normalize data across systems, about the difficulty of homogenizing data across internal and external inputs, the reality is that data that’s already in data format is sort of the easy part.
The harder part, and the next frontier, is taking the most unstructured data of all, spoken words, and structuring it.
In the parlance of the tech community, this is called natural language processing. Mixed with a dose of artificial intelligence (AI) and machine learning, companies across almost every industry are attempting to capture the conversations we have over the phone and in recorded meetings. Then they are trying to turn those conversations into structured data.
How does this help? Let’s look at a practical example in freight transportation. Most freight brokers to this day employ massive squads of people who pound the phones, securing capacity from carriers large and small. Even those brokerages with massive investments and deployments of optimization technology still rely on people.
The phone conversations are all slightly different, but also share significant characteristics. “Hey, how are you doing? I need a rate on this lane. That’s too much.”
This is where natural language processing and AI comes in. Using something called chatbots, technology companies are trying to eliminate the use of people to conduct these repetitive, low-value-add conversations.
A chatbot is a computer program designed to simulate conversation with human users. The idea is that these low-value transactional conversations are some of the least efficient things people in freight can do.
Last month, I met briefly with Parker Holcomb, the founder of a Chicago-based freight brokerage Fraight AI. Holcomb’s plan is simple: let companies on either side of his business use the systems they currently employ, and deploy language processing, AI, and machine learning to structure the data generated by conversations.
To be sure, Fraight AI is not the only one interested in such a path. IBM’s Watson is based around this same concept, the ability of a computer to act as a human would when asked a question.
IBM has recently made a strong push to sell Watson in supply chain management. IBM views Watson as a strategic asset within supply chains, a sort of butler that studies the language and behaviors of companies and then becomes an advisory entity.
Fraight AI has more modest ambitions at the start, befitting of a startup versus an established global conglomerate. Holcomb wants to create a brokerage that is more efficient internally and enables the shippers and carriers it works with to also become more efficient.
There’s also the price point. Watson is priced as a separate service, while Fraight AI’s engagements come embedded with its chatbot technology. That is the service, in other words. Even stripped out, Watson is likely to cost many multiples of what Fraight would charge its customers.
That’s not to say what Fraight AI is trying to do is easy. In some sense, it’s a question of time as Fraight’s engineering team “trains its models” to understand what is important about a conversation and what isn’t. That exercise takes time to execute – in AI and machine learning, information fed to a system exponentially increases its capability.
The use of bots is already permeating supply chain in various places. Programs that automate customer service functions, or last mile delivery coordination, are already in play.
The use of bots is already permeating supply chain in various places. Programs that automate customer service functions, or last mile delivery coordination, are already in play.
The use of chatbots backed by machine learning, on the other hand, IS not a well-developed concept in logistics yet. Which is what makes what Fraight AI and Watson are doing fascinating to watch.
As Murray Newlands, founder of enterprise chatbot platform Chattypeople, put it in a recent conversation with the tech journal Venture Beat: “Right now, text-based customer service solutions offering question-based responses are a big driver for bots, but I think that is going to change. On the user side, voice and proactive AI will change our interactions; bots will read out emails or heart rates and make recommendations about meetings or health suggestions. Enterprise chatbots will similarly pull in data from across businesses and external sources to make intelligent business management suggestions.”
Don’t believe in this revolution yet? Well how about this: Amazon announced in June it was making its Alexa speech recognition, text recognition, and conversational interaction capability available to outside developers through something called Amazon Lex.
Like when it made Amazon Web Services (AWS) a public commercial offering, Amazon providing a platform to allow external parties to increase the use of chatbots and AI-enabled language processing could help its usage explode in a short period of time.
At this point, it’s probably a matter of when smart chatbots are a fixture in logistics, not if. When you get into the office today, think about how many of your conversations are being captured, then think of all those conversations being structured and usable. It’s a pretty remarkable future that lies ahead.