DoorDash: Only Hundreds of AI Agent Orders vs 3M Daily Deliveries

Executive Overview

During its Q3 earnings call, DoorDash disclosed that external AI agents—including Muse, Instinct, and Fo—have generated only "several hundred" total orders, a striking contrast to the platform's average of 3 million daily deliveries in 2025. CEO Tony Shou emphasized that users still primarily order through the DoorDash app, while the company seeks to learn from agentic use cases and foster a healthy relationship with external agent developers. The disclosure provides hard numbers for the hype around agentic commerce, showing that AI-driven ordering remains a marginal channel. DoorDash is simultaneously building its own agent stack—text ordering, CLI, Ask DoorDash, and an MCP server—to control the user experience and capture orders wherever users are.

Deep-Dive Features & Architecture

DoorDash has rolled out a series of proprietary agents to capture orders across interfaces:

• Text Ordering Agent (September): Allows users to place orders via natural language text.
• CLI Agent (July): A command-line interface agent for programmatic or developer-oriented ordering.
• Ask DoorDash (June): A chatbot enabling orders through text and image commands.
• MCP Server: Implements the Model Context Protocol to let other chatbots place orders, targeting enterprise use cases.

This dual strategy—building internal agents to control the user experience and opening external channels to meet users where they are—reflects DoorDash's cautious but proactive approach to agentic commerce. CEO Tony Shou noted, "You don't want to fight gravity, but at the same time you don't want to be delusional," signaling adoption without overestimating the technology's maturity.

Benchmark & Competitive Performance

With only several hundred agent orders against 3 million daily deliveries, agentic orders represent less than 0.01% of total daily volume, confirming that agentic commerce is in a very early stage. Competitively, Amazon blocks personal AI agents from accessing its content, while Walmart opens its doors to them to attract business from rivals. DoorDash occupies a middle ground: engaging external agents cautiously while building its own. This positions the company to learn from early agentic use cases without ceding control of the customer experience.

Industry Impact & Enterprise Adoption

For developers in the Arab world, this announcement suggests that investing in Arabic-language ordering agents remains exploratory, given that global agentic orders are only in the hundreds. Development costs are relatively low now, but direct commercial returns are limited until the ecosystem matures. The lack of details on DoorDash's Arabic support in its agents leaves an opportunity for regional developers to build local solutions that master Arabic dialects and integrate with regional delivery platforms. Meanwhile, the MCP server opens enterprise integration possibilities, allowing businesses to embed DoorDash ordering into their own chatbots and workflows.

Conclusion

DoorDash's Q3 disclosure provides a reality check for agentic commerce: external AI agents account for only several hundred orders versus 3 million daily deliveries. Yet the company is investing in its own agent stack and an MCP server, betting on a future where AI agents play a larger role in ordering. For now, the data shows that consumers still prefer the native app, and the agentic channel remains a learning ground with future advertising potential. As the ecosystem matures, DoorDash's balanced approach—embracing agents without overestimating them—may prove prudent.

Media Source: TechCrunch AI | Fact Verification & Analysis: AI Tools Oasis