DoorDash reports only several hundred orders from AI agents

The reality of AI commerce

The tech industry is currently buzzing with the idea that AI agents will soon handle the majority of online commerce. These agents are software programs designed to perform tasks on behalf of humans, such as browsing websites and making purchases autonomously. However, recent data from DoorDash suggests that this shift is still in its very early stages.

During its third-quarter earnings call, DoorDash disclosed that it has only received "several hundred" orders from third-party agents. These agents include platforms like Muse, Instinct, and Fo. When compared to the company's total volume, these numbers represent a very small slice of the overall delivery market.

As TechCrunch reported, the company handles an average of 3 million deliveries every single day. This vast difference between the total volume and the agent-driven volume indicates that the technology has not yet reached a tipping point for mass adoption. Most consumers are still choosing to navigate the app themselves rather than letting an AI do it for them.

A balanced view from leadership

DoorDash CEO Tony Xu addressed the discrepancy between the hype and the data during an analyst Q&A session. He described the current landscape of AI agents as a dynamic and evolving area. Xu cautioned against being overly optimistic about the immediate impact while also acknowledging the potential for future growth.

Xu used a metaphor about gravity to explain his stance on the technology. He noted that you never want to fight gravity, but you also don't want to be delusional. This suggests that while the company recognizes the trend, they are not expecting an overnight takeover of their platform by autonomous software.

Despite the low current numbers, the company is not ignoring the trend. DoorDash expressed a desire to learn more about how people use these agents and wants to build a healthy relationship with the developers who make them. They are looking for ways to integrate with these tools rather than simply blocking them from accessing the platform.

DoorDash's internal agent suite

While third-party agents are currently making a small dent in the market, DoorDash is actively building its own suite of agentic tools. The company has released several different types of interfaces to help users place orders more easily. These tools aim to bridge the gap between human intent and the final transaction.

In June, the company launched a chatbot called Ask DoorDash, which allows users to place orders using simple prompts and photos. This was followed by a CLI agent in July, which provides a command-line interface for interactions. In September, they debuted a text-to-order agent to further simplify the process for customers.

To support corporate environments, DoorDash also released an MCP. This is a protocol that enables ordering from other chatbots specifically for business use cases. By creating these internal tools, DoorDash is positioning itself to be a primary destination for any AI that eventually gains the ability to make purchases.

The battle for agent access

The broader landscape of agent-based ordering is currently defined by a lack of consensus among major retailers. Websites are still struggling to decide whether they should allow AI agents to access their content and complete transactions. This creates a fragmented experience for users who want to use autonomous tools to shop across different brands.

Different companies are taking opposite approaches to this problem. Amazon has reportedly been blocking personal agents from interacting with its services. This move suggests a desire to keep users within their own ecosystem and maintain control over the direct relationship with the customer.

In contrast, Walmart has taken a different path by laying out a red carpet for these agents. The retail giant is actively trying to attract agent-based traffic to capture business from its competitors. This competition will likely determine which platforms become the default destinations for AI-driven commerce in the coming years.

Future opportunities and ads

Looking ahead, DoorDash sees a significant opportunity for growth within the agentic space. Even if the volume of orders is currently low, the company believes there is room for new types of revenue streams. One specific area of interest is the possibility of advertising within these third-party agents.

If AI agents become the primary way people discover and purchase products, the way brands reach customers will have to change. DoorDash is looking for ways to ensure their services are prioritized when an agent is searching for a delivery option. This could lead to a new model where brands pay to be the "recommended" choice by an AI.

The evolution of this space remains to be seen as the technology matures. As more agents become capable of handling complex tasks, the interaction between platforms and software will become more complex. For now, the data shows that the human-led app experience remains the dominant way people get their food delivered.