Amazon is introducing its new AI Shopping Guides feature. It promises to help shoppers better search through the e-commerce, giant’s virtual marketplace. It would also help sellers by working with Amazon’s Rufus chatbot.

How will Amazon AI Shopping Guides help consumers?

Amazon has been increasingly infusing the power of Generative Artificial Intelligence (Gen AI) in its multiple online platforms. The AI Shopping Guides are the latest introduction to online shopping on Amazon’s online marketplace.

Amazon claims the new feature will help out during the product research part. Specifically speaking, the feature will try to assist shoppers looking for a product by leveraging generative AI models. Amazon has mentioned that it intends to make “personalized companions” for shoppers using Gen AI.

AI Shopping Guides will appear after a shopper performs a standard search for an item. It is important to note that Amazon hasn’t added Gen AI shopping guides for all product categories yet.

Designed in the form of a guidebook, the Amazon Shopping Guides will reportedly offer an AI-curated collection of information about the products that buyers search for. This guidebook will contain basic information about the product category and charts comparing different specs

Amazon has been summarizing reviews in a single paragraph using Gen AI. The company will add this information to the AI Shopping Guides guidebook.

How will the new AI feature help sellers?

Amazon recently introduced the Rufus AI chatbot. It offers AI-powered research and recommendations. This AI chatbot is more geared towards promoting the products of sellers who advertise on Amazon.

The new AI feature would augment Rufus. It will reportedly help create guides from the information Amazon has about the products available for sale. Interestingly, it will also allow consumers to rewrite the descriptions provided by sellers.

The new AI Shopping Guides could help improve and simplify product descriptions without needing any significant intervention from sellers. Although unconfirmed, this feature might also partially take over the duties of the “Contact Seller” feature.

As the latest AI feature evolves, the e-commerce giant plans to collate and embed information from other web platforms too. Specifically speaking, Amazon might seek permission from consumers if they wish to personalize the recommendations by using data from Gmail and YouTube.

Meta will fix “mistakes” in how Threads enforces its rules after days of complaints about the company’s handling of content moderation on the service. In , Threads head Adam Mosseri said the company had already made some changes to address issues that have cropped up.

Mosseri’s comments come as Threads users have been increasingly vocal about Threads’ seemingly aggressive, and sometimes bizarre, moderation decisions. In one , a number of users reported that their accounts had been penalized for using the word “cracker” or “saltines.” Mosseri didn’t explain exactly why these types of mistakes occurred, but said that one of the company’s internal tools “,” which prevented human reviewers from seeing “sufficient context” about the posts they were moderating.

“For those of you who’ve shared concerns about enforcement issues: we’re looking into it and have already found mistakes and made changes,” Mosseri wrote. “Most prominently, our reviewers (people) were making calls without being provided the context on how conversations played out, which was a miss. We’re fixing this so they can make the better calls and we can make fewer mistakes. We’re trying to provide a safer experience, and we need to do better.”

Content moderation isn’t the only issue that’s rankled Threads users in recent days. Earlier this week, Mosseri also promised that Threads was working on a fix to bring “under control” on the service, following widespread complaints.

As more enterprise organizations look to the so-called agentic future, one barrier may be how AI models are built. For enterprise AI developer A121, the answer is clear, the industry needs to look to other model architectures to enable more efficient AI agents. 

Ari Goshen, AI21 CEO, said in an interview with VentureBeat that Transformers, the most popular model architecture, has limitations that would make a multi-agent ecosystem difficult.

“One trend I’m seeing is the rise of architectures that aren’t Transformers, and these alternative architectures will be more efficient,” Goshen said. “Transformers function by creating so many tokens that can get very expensive.” 

AI21, which focuses on developing enterprise AI solutions, has made the case before that Transformers should be an option for model architecture but not the default. It is developing foundation models using its JAMBA architecture, short for Joint Attention and Mamba architecture. It is based on the Mamba architecture developed by researchers from Princeton University and Carnegie Mellon University, which can offer faster inference times and longer context. 

Goshen said alternative architectures, like Mamba and Jamba, can often make agentic structures more efficient and, most importantly, affordable. For him, Mamba-based models have better memory performance, which would make agents, particularly agents that connect to other models, work better. 

He attributes the reason why AI agents are only now gaining popularity — and why most agents have not yet gone into product — to the reliance on LLMs built with transforms. 

“The main reason agents are not in production mode yet is reliability or the lack of reliability,” Goshen said. “When you break down a transformer model, you know it’s very stochastic, so any errors will perpetuate.”

Enterprise agents are growing in popularity

AI agents emerged as one of the biggest trends in enterprise AI this year. Several companies launched AI agents and platforms to make it easy to build agents. 

ServiceNow announced updates to its Now Assist AI platform, including a library of AI agents for customers. Salesforce has its stable of agents called Agentforce while Slack has begun allowing users to integrate agents from Salesforce, Cohere, Workday, Asana, Adobe and more. 

Goshen believes that this trend will become even more popular with the right mix of models and model architectures. 

“Some use cases that we see now, like question and answers from a chatbot, are basically glorified search,” he said. “I think real intelligence is in connecting and retrieving different information from sources.”

Goshen added that AI21 is in the process of developing offerings around AI agents.

Other architectures vying for attention

Goshen strongly supports alternative architectures like Mamba and AI21’s Jamba, mainly because he believes transformer models are too expensive and unwieldy to run. 

Instead of an attention mechanism that forms the backbone of transformer models, Mamba can prioritize different data and assign weights to inputs, optimize memory usage, and use a GPU’s processing power. 

Mamba is growing in popularity. Other open-source and open-weight AI developers have begun releasing Mamba-based models in the past few months. Mistral released Codestral Mamba 7B in July, and in August, Falcon came out with its own Mamba-based model, Falcon Mamba 7B.  

However, the transformer architecture has become the default, if not standard, choice when developing foundation models. OpenAI’s GPT is, of course, a transformer model—it’s literally in its name—but so are most other popular models. 

Goshen said that, ultimately, enterprises want whichever approach is more reliable. But organizations must also be wary of flashy demos promising to solve many of their problems. 

“We’re at the phase where charismatic demos are easy to do, but we’re closer to that than to the product phase,” Goshen said. “It’s okay to use enterprise AI for research, but it’s not yet at the point where enterprises can use it to inform decisions.”