Walmart Challenges Amazon with ChatGPT Shopping, Anthropic Targets Enterprise Workflows with Skills

ChatGPT's Instant Checkout brings Walmart to 800M users, Claude's Skills let enterprise users build custom workflows

Walmart Partners with OpenAI for ChatGPT Shopping

Walmart announced a partnership with OpenAI, allowing customers to complete purchases from Walmart directly in ChatGPT. Users can now ask ChatGPT for product recommendations, check inventory, and complete purchases without leaving the chat interface.

The integration leverages ChatGPT’s 800 million weekly users, giving Walmart access to a massive distribution channel. For OpenAI, this deal brings another major partner for their Instant Checkout feature, which already has Shopify & Etsy. Adding another vendor helps commoditize their e-commerce compliments, enabling OpenAI to set up a bidding system for all these vendors in the future.

Similar to Google with AdWords, the more vendors OpenAI can bring into Instant Checkout, the less power any of them have over OpenAI, and the more each vendor will bid for an incremental customer.

But this move is also a win for Walmart, as they have a massive third-party marketplace like Amazon. The customers Walmart acquires through ChatGPT are likely less inclined to be regular Walmart customers, and more likely to have purchased on Amazon instead.

The big question is if ChatGPT users will end up buying on Walmart through Instant Checkout, or pivot back to Amazon after finding what they need. I wrote last week about how powerful defaults are, and for Walmart, they’re so far behind on e-commerce, they might as well try.

As for whether Amazon joins, they’re likely in no rush. So many e-commerce searches already start with them, and partnering with ChatGPT could cannibalize this, as Amazon customers may decide to search on ChatGPT instead of Amazon.

But as I shared about my slide story, this decision could keep Amazon from being considered for the long tail of e-commerce. ChatGPT could bring their users to Amazon for products that the user didn’t know existed on Amazon.

It probably makes sense waiting, and that’s likely what Amazon will do. If Walmart, Shopify, and Etsy increase share from Instant Checkout, nothing’s stopping Amazon from doing a deal with OpenAI later. Given their partnership with Anthropic, Amazon could do a trial with Claude first.

Doing it sooner, though, could move eyeballs away from their search interface. This would not only cost them margin (paying a cut to OpenAI), but also advertising revenue from all of the ads they show.

Anthropic Targets Enterprise Workflows with Skills for Claude

Anthropic released Skills for Claude, a new feature allowing users to create reusable workflows that the AI can execute autonomously. Skills enable Claude to perform multi-step tasks like data analysis, report generation, or code review with a single prompt.

The feature targets enterprise users looking to automate repetitive workflows, positioning Claude as more of a customizable AI assistant. Users can build Skills through natural language instructions, with Claude learning the specific steps and applying them consistently across future tasks.

Their video shows an example of creating a skill, where they upload brand guidelines and other resources to a Zip file.

Once uploaded, these skills are referenced going forward. They then show Claude building a new Google Slide deck from scratch referencing these guidelines.

The big question, and ultimate challenge, will be ensuring Skills maintain accuracy and don’t introduce errors at scale.

Just looking at the example Claude shared, how well will Anthropic follow your brand guidelines? If users have to spend just as much time validating that guidelines were followed on every single slide, the output must be novel, otherwise the user may not have saved much time.

In other words, if the user isn’t really saving time generating content because that content must be validated, perhaps Skills can save time by brainstorming ideas that the user would not have considered.

The real test is adoption. Will enterprises invest time building Skills, or will they stick with existing workflows? Enterprises could wait until the product is more reliable, and who knows where that threshold is.