Fixing Robotaxi Custody for Mass Markets | Analysis by Brian Moineau

TL;DR

  • Uber charges a $15 fee for lost‑item returns, and its 2024 Lost & Found Index still lists phones as the top lost item—facts that frame the real AV challenge: industrializing the last 30 seconds of a ride across doors, trunks, and handoffs. [1][5]
  • Waymo rides have been bookable inside the Uber app in Phoenix since 2023, which makes Uber a multi‑party broker where custody and claims policy matter as much as the driving stack. [3]
  • To become the biggest robotaxi broker by 2029, Uber must make custody bulletproof at high‑throughput venues like Phoenix Sky Harbor (PHX) and SFO, where airports publish strict curbside SOPs that won’t forgive sloppy handoffs. [7]

What the source said

Uber’s Lost & Found Index (2024) highlights the usual parade of oddities and confirms that phones remain the most commonly lost items across its network—a reminder that misplacement is a stable human behavior, not a novelty. The help center sets a $15 rider fee for item returns, a standardized touchpoint that now extends to autonomy pilots. [5][1]

On the AV front, Uber and Waymo announced in May 2023 that Waymo’s robotaxis would be available in the Uber app, starting in Phoenix and expanding from there; that integrated booking step turns Uber into a front‑door for AV rides it does not operate. [3]

Waymo’s consumer support flow directs riders to report lost items through the Waymo One app, after which support coordinates retrieval—typically via depot intake or a scheduled return, which differs operationally from a human driver turning around. [6]

Why it matters

Three named stakeholders carry exposure. Uber, the broker, owns first‑line support and refunds, and must normalize custody across partner fleets, depots, and curbs in cities like Phoenix and Los Angeles by year‑specific playbooks rather than ad hoc chats. If it fails, complaints pile up in the same channels that drive MAUs. [1][3]

Waymo, the operator, owns curbside “edge‑case hospitality”—trunk confirmations, door interlocks, and remote assistance—already staffed by Fleet Response Specialists noted in Waymo’s safety reports. A clean trunk‑close at PHX, Chase Field, or Footprint Center is table stakes for brand protection. [8][7]

Airport and city authorities, from PHX to the California Public Utilities Commission (CPUC), publish and enforce precise pickup/drop‑off SOPs and AV permit conditions; unresolved custody at curbs can rapidly become a regulatory finding rather than a customer‑support ticket. [7][10]

Original analysis

Robotaxi lost and found: the underrated economics

The consensus: Lost‑and‑found in AVs is a quirky culture story.
The contrarian read: It’s an exception‑handling P&L story where small, frequent failures create real broker costs and reputational drag.

  • Evidence 1: Uber’s flat $15 lost‑item fee is an intentional cost anchor; the moment AV returns trigger courier legs and depot touches, that anchor shapes routing logic and SLAs. [1]
  • Evidence 2: Waymo publicly documents remote assistance roles that resolve on‑scene anomalies; those humans will adjudicate handoffs and custody questions that a Level‑4 system can’t “sense,” especially amid crowds leaving Chase Field after a 2024 Diamondbacks home game. [8]
  • Evidence 3: Airports publish curb management rules that already bind human ride‑hail; AVs inherit those timing windows and signage constraints, which forces precise, telemetry‑driven custody steps rather than vague “we’ll call you” promises. [7]

Back‑of‑envelope: courier costs scale faster than you think

Knowns:

  • Uber reported roughly 9.4 billion trips in 2023. [2]
  • Lost‑item return fee = $15. [1]

Scenario math (assumptions stated):

  • If AV trips on Uber reach 0.5% of total by 2026 and 0.05% of those AV trips require a couriered return, then yearly pass‑through courier fees ≈ 9.4B × 0.5% × 0.05% × $15. [1][2]
  • Work: 9,400,000,000 × 0.005 × 0.0005 = 23,500 AV courier events; 23,500 × $15 = $352,500 in courier fees before support labor, depot handling, or refunds. [1][2]

Interpretation: Even with conservative penetration and exception rates, six‑figure courier pass‑through arrives quickly; that is incentive to add AV‑native interlocks (door‑hold and trunk‑confirm) that shave exception rates by basis points.

A simple 2x2 for custody maturity

  • X‑axis: Vehicle autonomy capability (supervised → driverless).
  • Y‑axis: Custody instrumentation (manual → telemetry‑confirmed).

Quadrants:

  • Supervised × Manual: Safety driver checks cabin; cheap but unscalable.
  • Supervised × Telemetry‑confirmed: Safety driver plus app prompts; training ground for SOPs.
  • Driverless × Manual: Support‑chat roulette; slow, inconsistent, brand‑risky.
  • Driverless × Telemetry‑confirmed: Door/trunk sensors, dwell‑time rules, positive rider confirmation, depot scan‑ins; the only quadrant that scales to airports and stadiums.

Historical analogue: UPS introduced the DIAD handheld in 1991 to scan parcels at every handoff, which crushed dispute rates by logging custody at each node; robotaxis need the DIAD equivalent for riders’ belongings. [9]

Named‑stakeholder breakdown

  • Uber: Ship in‑app interlocks such as “trunk confirm” and “door hold” tied to curb geofences at PHX and SFO; route unresolved cases to partner depots with clock‑started SLAs. Miss this, and refunds and airport fines eat margin; hit it, and custody becomes a defensible broker moat. [7]
  • Waymo: Tighten curbside SOPs where rides are bookable via Uber in Phoenix, and publish artifacted logs (door state, trunk actuation, dwell time) to shrink dispute windows to hours, not days; lean on Fleet Response Specialists to clear edge cases. [3][8]
  • Regulators and venues: CPUC and airport authorities will codify trunk/door dwell minimums and return windows once volume rises; early compliance wins lift‑and‑shift across markets in 2025–2027. [10][7]

This is why “weird stuff left in robotaxis” is not a sideshow; it is the probe we can use to measure whether AV networks deliver hospitality, not only autonomy.

What others are missing

The angle: chain‑of‑custody will harden into a platform standard with telemetry primitives—door‑open states, trunk release logs, rider proximity pings, depot intake scans, and venue geofences—rather than a soft help‑center script. Waymo already runs remote assistance roles, and Uber already standardizes return fees; wiring those facts into a cross‑partner custody API lets the broker mandate positive confirmation before any trunk closes at PHX or Chase Field, pushing exception rates down and making partner onboarding a policy load, not a bespoke ops sprint. [1][8][7]

What to watch next

  1. By December 31, 2026, Uber will publicly list at least one additional U.S. city beyond Phoenix where AV rides are bookable in‑app (via press release, newsroom post, or investor deck), and third‑party outlets will confirm it. [3]

  2. By June 30, 2026, Uber’s Help Center or Newsroom will publish an AV‑specific lost‑item workflow that differs from the human‑driver flow (mentioning depot intake or partner coordination), with a dated update page. [1]

  3. By March 31, 2027, at least one major U.S. airport (PHX, SFO, or LAS) will publish an AV curbside SOP that includes explicit rules on luggage handling or dwell times for driverless vehicles, available on the airport’s official site. [7]

My take

The denture jokes distract from the 1991‑style DIAD lesson: custody is a data problem that decides unit economics. Uber and Waymo already have the building blocks—flat fees, remote assistance, and instrumented vehicles—and the broker who turns those into a custody standard will set the industry spec by 2027. If they do, “largest robotaxi broker by 2029” reads less like swagger and more like a normal consequence of better exception math.

Sources

  1. Uber Help Center — Lost items and the $15 return fee — Establishes the standardized rider charge and baseline flow for lost‑item returns.

  2. Uber Investor Relations — Q4 2023 results (press release/letter) — Provides the ~9.4 billion 2023 trip count used in the back‑of‑envelope math.

  3. The Verge — Waymo and Uber partnership announcement (May 2023) — Confirms that Waymo rides are available inside the Uber app in Phoenix.

  4. TechCrunch — Waymo expansion to Austin (August 2023) — Documents planned AV service areas beyond Phoenix and the cadence of city additions.

  5. Uber Newsroom — 2024 Lost & Found Index — Confirms that phones are the top lost item and provides context on item types and seasonality.

  6. Waymo One Help Center — Lost and found process — Describes how riders report and retrieve items via support and depots rather than