Delivery-based grocery retailer Teddy is using ChatGPT to grow further and spend less.
Delivery-based grocery retailer Teddy is using ChatGPT to grow further and spend less.
Artificial intelligence is propelling a Kiwi grocery startup to new heights, letting them compete with big industry players and saving hundreds of thousands on staff.
Teddy co-founder Ricki Taiaroa said the delivery company is using chatbot ChatGPT for online grocery recommendations, running code and warehouse operations.
Taiaroa said local startups now have “the opportunity to establish a business with fewer resources than you needed in the past”.
“It enables us to more effectively compete,” he said.
“The name of the game is building a competitive business so that in [a] year’s time, we’re able to compete with the bigger guys and we’re an established brand. Technologies like [ChatGPT] enable us to grow sustainably.”
He said Copilot alone is doing the work of one to two developers.
“It is running a lot of code. It’s a different way of programming, but it’s a productivity booster. It’s quite powerful,” Taiaroa said.
The co-founder said product recommendations are Teddy’s main focus in using ChatGPT: “When you’re shopping around in the app, it will look at what you’re browsing and give you better recommendations.”
“In a previous role, if we were to buy a recommendation engine off the shelf, you’re talking hundreds of thousands of dollars, if not probably over half a million,” he added.
“We’ve been able to build it with some text prompts and ChatGPT for a few hundred dollars.”
He said the company’s convenience-based business model means predicting customer needs is essential.
“Before technology like GPT, you would have to build a machine learning model that would be based on historical sales data. It would be very hard to build a model that can understand products at the depth that GPT can understand things.”
Teddy founders Ricki Taiaroa (left) and Chaz Savage.
Taiaroa said the technology is “hard to match” compared with manual software development.
He said the team are now looking at adding automated product tagging to their AI basket.
“When we bring new products into the warehouse, it’d be able to tag and categorise those products in the warehouse for us” bringing more savings in warehouse operations, he said.
Chief technology officer Dion Yiw said while ChatGPT gives business insight around consumer behaviour, the technology is “already quite powerful” after being launched only a few months ago.
“I think pretty soon we’ll have these sorts of software agents that can understand every part of a business, about the business cycle, your customers, all your products,” Yiw said.
Taiaroa added: “We can use it to enhance our warehouse operations, it’s already doing a bunch of our coding for us. So when you put all that together, we can replicate a lot of the capabilities that a bigger team would have.”
The Mind Lab and tech Future’s Lab founder and chair Frances Valintine said AI has the potential to change the way developers work. “The ability to automate highly repetitive tasks can easily be written using AI. This speeds up project time and also removes human error.”
She said AI can also fast-track quality assurance testing, debugging and user testing for businesses.
Teddy's Auckland warehouse.
“Generative AI is ideally suited to code recommendations in the same way Grammarly helps people to write.”
She said, “The recommendations are very likely to be helpful, fast and often will improve on what was there in the first place, but engineers will still need to validate and oversee AI-generated code at least for the immediate future.”
When asked if AI curbs the need to hire developers, Valentine said junior or entry-level developers will be challenged and in many cases replaced with AI.
“Over time, further AI advances will increasingly provide more effective solutions for the coding aspects of software development.”
She added, however, that there is still “considerable time” before generative AI can problem-solve and create solutions in the same way senior software engineers can.
“I wouldn’t want to put a timeframe on this as we are seeing improvements and significant new entrants in the AI space emerging each week.”
She said the risk in a slow adoption of AI will be a disadvantage to a business reluctant to use the technology.
“There are also many advantages of having AI undertake tasks where there is a shortage of suitable local talent. In the current market, very few companies with development teams are able to hire as many staff as they would ideally like due to talent shortages and increased cost of employment.”
University of Auckland senior marketing lecturer Marilyn Giroux said: “ChatGPT can considerably help smaller businesses in mundane, repetitive and routine administrative tasks to save time, money and improve efficiency.”
However, she said, businesses should be careful about using AI in marketing.
“We still need to investigate how humans react to AI. Recent research suggests that individuals respond to humans differently than AI.”
Her research shows consumers are more likely to behave unethically when dealing with artificial intelligence than humans.
“We attribute different intentions to AI than humans, like responsibility, but also exhibit divergent moral intentions and behaviours when dealing with AI,” Giroux said.
She said ChatGPT’s current function can’t replace the human factor in marketing.
“Businesses still need to validate the work, research and ideas as it can generate incorrect information and biased content leading to misinformation.”
She said companies need human experts who have more knowledge about a business to keep it successful and relevant to their audience.
