Since we launched AI Price Hunter, the price-discovery AI shopping agent, early this year, we continue to make engineering and product innovation. We launched AI Price Hunter v2 in April. I like to share the exciting improvements we have made in v2.

First, we improved agent reliability and accuracy. For the state-of-the-start AI shopping agent, it is a challenge to complete the increasingly complex and multi-step task all the time. We keep optimizing our workflow and test different foundation models. We have increased our SPU-level task completion rate to 80%. At the same time, we are optimizing our algorithm to improve hunting accuracy continuously.

Second, we enrich our data sources. The underlying data is always critical for any AI shopping agent. Besides leveraging our accumulated data set with tens of millions of merchandising items, we are also using external data sets. We qualitatively and quantitatively evaluate each external data set for our AI price hunter’s needs and spend a lot of effort curating the data sources. It is a time-consuming experiment and evaluation process. Hopefully, the AI price hunter can be the go-to price discovery agent for any consumers who are interested in price discovery in sports and fashion. Of course, the underlying AI price hunter can be adjusted for other shopping categories.

Third, we improved product design and provided API support for any merchant who is interested in AI price hunter service. We are happy to announce that CloudStore AI has included AI price hunter API. We have provided two package options, Hunter and Hunter Pro. Contact us for more information

Fourth, we launched the interactive feature. When people talk about AI agents, people often talk about automation and intelligence. Interaction is much less discussed. But interaction with human or other AI agents is a very important part for AI agents for complex and high-risk tasks such as purchasing a sneaker. AI Price Hunter started this interactive journey early. See the screenshot below as a case of the AI Price Hunter used in BeyondStyle. The user can submit her hunting request to the AI price hunter. The AI price hunter will put their request in a queue. When the hunting is over, the AI price hunter will notify the user of the result. We call it the one-shot interactive version of AI price hunter.

It is our decade work of transforming global commerce in making. Ten years ago, we embarked on a bold journey to revolutionize global commerce by making it automatic, intelligent, and interactive. Our vision has been clear from the start: 1-2-3 — automatic, intelligent, and interactive. Over the past decade, we’ve made remarkable strides:

• For 10 years, we’ve been automating global commerce, streamlining processes to make trade seamless through RPA (Robotic Process Automation) technology.
• For 8 years, we’ve been infusing intelligence into global commerce, leveraging cutting-edge technology to drive smarter decisions such as smart product recommendation.
• In 2024 , we’re taking a significant leap forward with the development of AI Shopping Agents, bringing a new level of intelligence (LLM technology in particular) to the shopping experience.
• And now, we’re just stepping into the future of interactive commerce by releasing a one-shot AI price hunter.

We’re incredibly proud of how far we’ve come and excited about the road ahead. Here’s to continuing our mission of transforming global commerce for the better!

We are also looking for collaboration with other AI shopping agents to build the agentic AI together, the future of the multi-agent world. Contact us if you are working in AI shopping agent space and looking for partnership.