Uber Data Reveals AI Agent Adoption Is Negligible
AI-generated image · Bay Street Wire
Despite industry hype, actual transaction volumes for AI agents remain strikingly low at major platforms like Uber and DoorDash.
The promise of an 'agentic' web—where AI handles real-world tasks on behalf of users—is facing a stark reality check in transaction data. During a third-quarter earnings call, Uber revealed that during its busiest week in October (the week of Halloween), only four rides were ordered via AI agents, as TechCrunch first reported.
While Uber Eats saw slightly more activity, the volume remains marginal. Uber CFO Balaji Krishnamurthy told analysts that out of approximately 320 million total trips and orders during that same Halloween week, only 500 Uber Eats orders were placed by AI agents. Uber CEO Dara Khosrowshahi acknowledged the "very limited traction" so far, noting that the efficiency of the company's own app—where booking a ride takes 50 seconds and re-ordering food takes 90 seconds—means AI agents may not be offering a significant time-saving advantage.
Uber is not alone in this trend. TechCrunch reports that DoorDash also disclosed on a recent earnings call that it has received only a few hundred orders from AI agents to date, as the vast majority of customers continue to use the DoorDash app.
These figures contrast with significant market momentum. Meta reported that its Muse app reached 10 million users last month, becoming the top app in the U.S. App Store, with many users utilizing it for business transactions. Additionally, the startup Instinct recently raised funding at a $10 billion valuation. While major players including OpenAI, Google, Anthropic, and SpaceXAI (Grok) are deploying agents, the current data suggests a wide gap between user acquisition and actual API execution in the real world.