AI finds a place in ag customer service

Quick, good answers to basic technical questions at dealerships are around the corner through the power of artificial intelligence (AI). 

Last November, access to cloud-based AI became real when OpenAI introduced ChatGPT. Now large language model (LLM) algorithm variations are being offered by companies such as Anthropic, Cohere, Databricks, Google AI, Meta AI, Salesforce, and more. 

Each model is loaded with the data and the language ability to answer questions. 

Technicians at service desks are the traditional human in the loop between farmers and their machines, but an “AI button” on the keyboard to bring up instant answers would be of great value.

That opportunity is at the doorstep, says Larry Hertz, North American Equipment Dealers Association vice president for Canada. He knows the 4,000-member association is struggling to provide the staffing and phone lines to answer customer questions in busy seasons.  

“In any state or province, most dealers are hard-pressed to have enough technicians,” Hertz says. “If they can make the technicians more effective or efficient, that’s better for farmers and the industry.” 

What dealers do have at every service desk are rows of documentation (service, operator, parts, and technical manuals) and other supplies for answering questions. Available nearby is Internet access with a browser. 

A solution for the technician shortage took a giant step forward earlier this year when two companies turned AI on the available data. An Indiana company began setting up AI support for the service desk to make all that existing documentation easily available. About the same time but on a larger scale, a Canadian consulting company began developing an app for an LLM system offering a mechanism for dealerships to integrate data resources easily with AI models.

Headsight AI model

Headsight Harvesting Solutions, Bremen, Indiana, emerged in the 2000s with a solution for controlling long corn headers accurately from the drivers’ seat of a harvester. Today, the company, part of AGCO, has a fleet of products for automatic height control and row guidance.

The Headsight technical support line recorded 2,000 hours of questions and answers in 2022. Product support “is something we do really well already,” says Bernardo Ferreira, general manager. “We try to improve the life of farmers in every interaction.”

ChatGPT impressed Ferreira and his leadership team, who wondered how they could use it to help farmers the most. Their answer: Improve the product support team performance. 

In April, they started loading digital transcripts of the 2,000 hours of company Q&A into an AI model. Then they uploaded every page of every manual or piece of documentation (about 5,000 documents), and their own method of organization, making sure to maintain privacy by not feeding any personal information into the AI databank. 

It took nine weeks for more than 10 people to develop a working version that could answer a technical question, and it was put into service in mid-September. 

“The model is already being used by our product support team,” Ferreira says. “We are seeing three gains: increased percentage of getting it right the first time, increased percentage of first operator finishing the call without support from a more experienced team member, and faster response time.”

The project is a work in progress. “It’s not about cutting costs. It’s about serving people. In time, we’re expecting it to show in sales volume and to raise the performance of all product support employees to the same level as our best experts,” he says. 

“We are not able to measure these gains yet, but the qualitative assessment of our team has been really good,” Ferreria adds. “We expect gains of more than 10 percentage points in each one of these areas.”

VisorPro AI Solution

AGvisorPro was formed in November 2018 in Calgary, Alberta, by cofounders Robert Saik, former head of Agri-Trend Technology, and Patrick Walther, a Swiss-based agronomist. The company connects farmers with questions to experts and organizations with answers, according to the website (agvisorpro.com). 

When ChatGPT came along, Saik and Walther saw opportunity. If the AI model could be trained and if privacy could be protected, AI had the capacity to open the pinch point between questions and answers handled by dealerships. During planting, spraying, and harvesting, a single service desk commonly receives hundreds of calls a day. 

Saik and Walther brought in Canadian physicist and technologist Brock Moir as chief product officer to develop their solution. Moir has worked with top physics researchers on the Large Hadron Collider project and with top AI researchers at the Alberta Machine Intelligence Institute. 

About 10 dealership pilot partners from Saskatchewan and Montana signed on to use VisorPro LLM after the July 6 launch. Dealer members are assured of their own vault to protect their information while it resides in the OpenAI LLM cloud.  

Since then, Saik and Walther have addressed interested dealerships in Canada and the United States. 

“Our VisorPro AI technology is really resonating with equipment and precision ag dealers,” says Saik. “The pinch point we are addressing with VisorPro is an acute shortage of technical staff, their burnout, and their churn at the dealership level.”

Moir adds, “Wherever there are technicians who need to constantly refer back to specifications and technical manuals, there’s interest. People outside the farm equipment space [are] reaching out to figure how they can become part of it. It would be good for heavy industry, airlines, automotive, construction, forestry, even the RV industry.”

Regina, Saskatchewan-based Young’s Equipment sign-up was followed July 18 by Torgerson’s LLC of Ethridge, Montana. Both operate multiple CNH dealerships, including equipment available through Case IH, Case Construction, and New Holland. Since then, other dealers have joined. 

“Our partnership with AGvisorPro in launching VisorPro marks a significant leap forward in the industry,” says Sean Young, assistant general manager at Young’s Equipment. “The ability to streamline our support process using AI technology will revolutionize our customer service offering and set a new standard in the dealership landscape.

“We’re reshaping our approach to customer service,” he adds. “Through VisorPro, we’re increasing the efficiency of our technicians and ultimately supporting our customers better with immediate, accurate responses to their queries. We’re setting a precedent for efficiency, quality of service, and innovative thinking in our sector.”

Torgerson’s LLC CEO Brion Torgerson declared, “This is revolutionary. During our peak seasons, the pressure on our technical and service personnel is intense. The integration of VisorPro signifies a game-changing opportunity for Torgerson’s in terms of technician onboarding, training, and overall business operations. This speeds the service process up immensely.” 

A setup period is underway to get dealers ready for spring. “There is quite a bit of work to do to load their data and get them ready to take full advantage of VisorPro,” says Saik. “After that, training will be ongoing between staff and VisorPro. Each addition or correction to the dealer’s vault will make VisorPro more valuable.”