Using AI to Chat With Your Database

Keyboards May One Day Be No Longer Key to Business Communication
Getty Image of spoken AI
bgblue/Getty Images

[Find the latest in trucking technology: Explore this quarter's issue of iTECH]

Early adopters are using artificial intelligence to have daily conversations with their databases to mine business insights, freeing them up from keyboards forevermore.

The capability to talk with your database each morning to glean insights on how to further maximize your profits may seem like sci fi to some.

But for those in the know, chatting daily with their databases — as if they’re talking with Siri, Alexa, Google Assistant or a similar voice interface system — has become second nature.



“Conversational AI will help us get one step closer to making sense of our data,” said JT Peters, CEO of Truck Driver Power, a blog for truckers. “But we need to be cautious — not to trust it blindly.”

Image
Joe Dysart

Dysart 

Jane Jazrawy, CEO of online driver training provider Carriers­Edge, highlighted ways the freight transportation industry might utilize this capability.

“Extracting data in a report form or manually moving numbers to a spreadsheet is time and labor intensive, so any way to make it simpler would be advantageous,” she said. “For example, recruiting or HR should be able to find out trends in when and why people are leaving the company. And sales should be able to access information about customer history and driver issues during a pickup or delivery.”

In theory, voice-accessible systems can enable workers in virtually every department in your trucking organization. And they’re able to offer those workers quick access to that data — whether they are casually working on some research or need quick access to some insights before a meeting.

Image
Jane Jazrawy

Jazrawy 

Moreover, these systems enable workers to converse with company databases in a highly specific way, allowing them to drill down with follow-up questions and what-ifs to retrieve specific solutions for specific challenges.

Even better, that valuable access to data and analysis can be granted to all employees at a trucking company when warranted — not just data scientists and top executives.

Plus, getting your database to work with voice query can be relatively easy. Numerous AI-­powered voice query systems have already been designed to seamlessly interface with commonly used business intelligence systems such as Snowflake, ­Power BI, Pentaho Analysis and Microsoft SQL Server.

So if your trucking business is already using one of those BI systems, you should be able to easily convert your access to its data and analysis to voice.

Most voice access systems are also often extremely robust. No matter how much information your company currently has stored in its database, there’s a good chance the AI-powered voice query system can work with that data.

For example, an operations manager at a trucking business can naturally converse with a company database and drill down with questions such as:

  • How many trucks are overdue for maintenance, where are they located and how quickly can they be routed to maintenance facilities?
  • How do our sales for this June compare with June sales from the previous year, and what key factors are influencing the difference in those numbers?
  • Who are the top 10 drivers in our company right now in terms of overall performance, and what characteristics do they share?
  • What is your projection for total 2023 sales, and what are the top five factors influencing that projection?

In each case, you’ll never need to go near a keyboard to access this kind of data and analysis. Instead, you can just use your voice and get the answers spoken back to you — just like you’re talking to Siri, Alexa, Google Assistant or a similar voice interface system.

Plus, these same systems are also able to serve up all the typical charts, graphs and ­other visuals you’re accustomed to seeing with a keyboard-­driven business intelligence system.

Image
Warren Powell of Optimal Dynamcs

Powell 

“Using voice to enter commands — and then using screens to display lists and maps — seems like an unbeatable combination,” said Warren Powell, chief innovation officer at Optimal Dynamics, an AI-powered logistics management company.

Yet another beauty of AI-powered voice query is these systems have the ability to handle new types of questions on the fly.

The reason: AI-powered voice query is, by its very nature, designed to handle new types of questions, come up with an insightful answer and then auto-integrate that type of question into its system so that the next time around it’s better prepared to handle that kind of query.

Image
RoadSigns

Chad Crotty of DDC FPO Solutions discusses how fleets handle outsourced business services. Tune in above or by going to RoadSigns.ttnews.com.  

Plus, workers using these systems find they no longer need to go with their gut on a decision because the keyboard-­access company business intelligence system is too hard to use.

Instead, these employees simply walk in each morning and get instantaneous feedback from the company database throughout the day as if they’re consulting a trusted assistant.

Compare that immediate access to data and analysis to traditional employee interaction with a business intelligence system.

That kind of workflow — which is being increasingly perceived as antiquated — often requires employees to wait weeks or months for data scientists at the company to come up with a new kind of report they need.

As for data scientists at organizations that have opted for AI-powered voice query, they’re generally delighted to no longer be bogged down with an endless stream of requests for basic business reports from users from vir­tually every department in the organization.

Want more news? Listen to today's daily briefing above or go here for more info

Instead, these data scientists have been freed to engage in the kind of increasingly sophisticated analysis for the organization that they’d really rather be doing.

Of course, despite the myriad advantages of voice access to a database, there are some caveats to consider before going all-in.

You need to ensure, for example, that the AI system you’re considering offers you true access to actual analysis of the data in your system — and not simply easy answers to easy questions.

“If it is limited to only responding to simple requests like, ‘What are the dock hours for the ABC facility?’ or ‘What is the billing address for customer XYZ?’ it is not a big game ­changer,” said Chris Caplice, executive director at the MIT Center for Transportation & Logistics. “However, if it can respond to something like, ‘Tell me the customers who are more than 30 days late in paying and have a history of not paying for accessorials’ or ‘Which loads are at risk of being more than one day late?’ it can really improve operations.”

There’s also a good chance you may need to provide many workers training on how to question the database — especially workers who currently do not access your trucking company’s database directly.

At trucking firms with keyboard-access database and analysis systems, those workers generally rely on reports or dashboards created by a data scientist.

Image
Martin Cousineau of HEC Montreal

Cousineau 

“These systems are often sensitive to the way questions are asked,” said Martin Cousineau, assistant professor of logistics and op­erations management at HEC Montreal. “Asking the question in a slightly different way can often produce quite different results.”

There’s also a possibility that an AI voice query system that is not fully trained and customized for your trucking business may not understand — or may misinterpret — what your employee is asking.

Consequently, those less-than-optimal systems may respond with an answer that is off base or simply wrong.

Traditional database systems generally avoid these kinds of problems, given that company data scientists are charged with ensuring that the questions asked of a database are understood correctly by the AI system, Cousineau added.

Essentially, they serve as intermediaries between workers who need data and analysis and the system that provides those insights.

David Correll, co-director of MIT FreightLab, agreed there are dangers associated with offering inexperienced workers direct access to data and analysis.

“The quick and beautiful answers that computers provide can be deceiving,” Correll said. “Even the most sophisticated algorithm still suffers from the ‘garbage-in, garbage-out’ problem.”

Bottom line: Any system deployed organization­wide would need to be exhaustively tested and refined to ensure such mistakes do not occur.

Yet another danger in offering voice query companywide is security breaches.

Any AI-powered voice access system would need to be secured to ensure the AI knows who is accessing the system at any given time, according to Jazrawy of CarriersEdge.

The system would also need to be designed so that it can block people who do not have privileges on the system, she added.

Plus, company security officers would need to develop strong guardrails against unethical employees who might sell access to your data and analysis to a competitor, Jazrawy said.

“How do you control who is listening or recording?” she asked.

Granted, there are a number of potential pitfalls to consider when bringing a conversational database system onboard. But the numerous advantages offered by easy voice access to company data and analysis have inspired a number of early adopters to bring the AI tech in-house, just the same.

Vendors offering voice access to data and analysis include Genesys, Coruzant, Acuvate and Iovox. 

Joe Dysart is an internet speaker and business consultant based in Manhattan. Voice: (631) 233-9770. Email: joe@customtechadvisor.com. Web: www.dysartnewsfeatures.com.