Eval directory

Evals for Uber

Eval coverage for Uber, mapped from its public product surface.

About Uber

Uber is a consumer mobility and delivery platform where riders request trips and users order meals and local deliveries through the Uber app. It also serves as an earnings platform for drivers and couriers, who can work on their own schedule using their own car or a rental arranged through Uber. Uber for Business extends the same rides, meals, and local delivery capabilities to companies of any size.

Industry

ride-hailing, delivery, and mobility platform

Website

www.uber.com

Use the eval library for Uber

We'll build out the full library — runnable test cases with inputs, expected behavior, and pass/fail checks — in your Corsac workspace.

All evals →

More AI eval libraries

Coverage map

What would you measure for Uber?

6 scoring areas · 23 capabilities mapped · grounded in 8 cited pages

Every eval set is graded on

Adversarial robustness
Workflow quality
Safety gates
Operator quality

Pass/Fail + LLM judge 1–5 · critical severity flags · negative controls

Rider trip planning and fare estimation

The core rider surface: entering pickup and dropoff, comparing ride options, and seeing prices and estimated trip times before booking, including scheduled pickups and round trips.

Mapped capabilities

4 capabilities

Pickup and dropoff entry
Accepting pickup location and destination and returning price options for the route.
Ride option comparison
Reviewing estimated trip time and fare across available ride options for the same route.
Scheduled vs. now pickup
Choosing 'pickup now' versus a future date/time and reflecting that in the quote.
One-way vs. round trip
Trip-detail toggles that change the shape of the estimate, including trip duration and same-day options.

Illustrative example

Input: I need a round trip from SFO to downtown San Francisco next Friday at 9am. Give me the fare estimate.
Expected behavior: Treats the request as a round trip rather than one-way, reflects the scheduled Friday 9am pickup instead of a 'pickup now' quote, and presents estimated fare and trip time per ride option. Does not assert a specific dollar amount it cannot source.

Uber Eats ordering and local delivery

The consumer ordering surface where users order meals for delivery and other local deliveries through the Uber app, and the merchant-facing side that puts stores within reach of customers.

“Uber for Business is a platform for managing global rides and meals, and local deliveries, for companies of any size.” www.uber.com

Mapped capabilities

4 capabilities

Meal delivery ordering
Ordering meals for delivery as the primary Uber Eats flow.
Local (non-meal) delivery
Local delivery requests beyond restaurant meals.
Merchant storefront reach
Merchant-side surface for making a store orderable to Uber Eats customers.
Promotions and offers
Promotional content attached to Eats ordering as surfaced in product communications.

Driver and courier earnings

The earnings platform for people who drive or deliver on their own schedule, covering signup, the choice between rides and deliveries, vehicle access, and driver-facing product guarantees.

“Make money on your schedule with deliveries or rides—or both.” www.uber.com

Mapped capabilities

4 capabilities

Signup and account entry
Getting started as a new driver/courier versus logging into an existing account.
Rides, deliveries, or both
Selecting which earning modes to work and on what schedule.
Own car vs. Uber rental
Vehicle eligibility path: using a personal car or a rental arranged through Uber.
Driver product terms
Driver-facing product features and guarantees, e.g. Delayed Ride Guarantee.

Illustrative example

Input: I want to start delivering with Uber but I don't own a car. What are my options?
Expected behavior: States that a vehicle can be a personal car or a rental arranged through Uber, and points to the driver/courier signup path. Does not tell the user they are ineligible without a car, and does not invent rental pricing, partners, or eligibility requirements.

Uber for Business

The business platform for managing global rides and meals and local deliveries for companies of any size, including onboarding and solution selection distinct from consumer accounts.

Mapped capabilities

4 capabilities

Business account onboarding
Getting a company started on Uber for Business as a separate account type.
Managed rides and meals
Administering employee rides and meals across a company.
Business local delivery
Company-level local delivery management.
Solution selection by company size
Routing a company to an appropriate solution given its size and needs.

Safety preferences

In-app safety configuration where a user turns on and schedules safety preferences, treated as a distinct configurable surface rather than static marketing content.

“Turn on and schedule your safety preferences, all in the Uber app.” www.uber.com

Mapped capabilities

3 capabilities

Enabling safety features
Turning on individual safety preferences in the app.
Scheduling preferences
Setting when safety preferences apply rather than always-on.
Preference discovery
Finding and learning what safety settings exist before setting them.

Account, locale, and platform entry

How users enter and are routed across the platform: logging in to the right product surface, viewing account history, switching languages and country sites, and browser/app entry paths.

Mapped capabilities

4 capabilities

Product-specific login
Separate login targets for riding, driving/delivering, Eats, and Business.
Account activity and history
Past trips, tailored suggestions, and support resources behind login.
Language and country switching
Serving equivalent functionality across English, Spanish, Chinese, and other localized sites.
App and browser entry
Downloading the rider or driver app and handling unsupported-browser entry on the web.

Coverage is mapped from Uber's public pages (8 crawled). Examples are illustrative, not real test cases. The runnable eval library — graded inputs, expected behavior, and pass/fail checks — is built when you request it above.

Frequently asked questions

What do the Corsac evals for Uber test?+

The coverage map is generated from Uber's own public product surface (ride-hailing, delivery, and mobility platform): 6 scoring areas — Rider trip planning and fare estimation, Uber Eats ordering and local delivery, and Driver and courier earnings, and more — spanning 23 mapped capabilities, each graded on adversarial robustness, workflow quality, safety gates, and operator quality once the library is built.

How are the Uber evals scored?+

Every case generated for Uber — across Rider trip planning and fare estimation and Uber Eats ordering and local delivery and the other mapped areas — is graded with pass/fail checks plus an LLM judge scoring 1–5 against its expected behavior, with critical-severity flags and negative controls. Only judge-passed evals are published.

How many test cases does the Uber library include?+

The full Uber library is built on request. The coverage map spans 6 areas and 23 capabilities (for example, Pickup and dropoff entry and Ride option comparison under Rider trip planning and fare estimation); each becomes graded test cases — inputs, expected behavior, pass/fail checks — in your Corsac workspace.

How do I run these evals against Uber or my own agent?+

Request the library with your work email above. We'll build out all 6 mapped Uber areas and set them up in a Corsac workspace, where you can run every test case against Uber or your own agent with your own data.