Eval directory

Evals for Flip CX

Eval coverage for Flip CX, mapped from its public product surface.

About Flip CX

Flip is a Voice AI platform that automates inbound customer support phone calls for retail eCommerce, healthcare, and transportation brands. It handles call types such as returns and exchanges, subscriptions, appointment and ride scheduling, prescriptions, insurance checks, and order-by-phone rather than just order-status lookups. The company says it serves hundreds of brands and automates millions of calls per week.

Industry

customer support voice AI

Headquarters

New York City (NYC HQ, opened 2018)

Website

flipcx.com

Use the eval library for Flip CX

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 Flip CX?

6 scoring areas · 21 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

Retail eCommerce Call Handling

Automating the common inbound retail call types Flip positions as 'not just WISMO' — resolving intent end to end rather than reading back an order status.

“Automating up to 90% of calls for fleets, call centers, and medical transport coordinators.” flipcx.com

Mapped capabilities

4 capabilities

Returns and exchanges
Identify the order, determine eligibility, and drive the return or exchange to a stated outcome.
Subscription management
Handle subscription-related caller requests such as changes to an existing plan.
Product and store info
Answer product and store questions from brand-provided sources without inventing details.
Order by phone
Take an order over the phone, including capture and confirmation of what was ordered.

Illustrative example

Input: Caller: "Hi, I want to return the jacket I bought. I think it was back in March — the order was under my email." Brand return window is 30 days from delivery.
Expected behavior: The assistant looks up the order, determines the purchase falls outside the brand's 30-day return window, tells the caller the return is not eligible, and offers the brand's configured alternative or a handoff rather than starting a return.

Healthcare Practice Calls

Voice handling for providers, urgent cares, dental offices, physical therapy clinics, and veterinarians — beyond appointment booking, per Flip's healthcare positioning.

“Modern Voice AI built for healthcare providers, urgent cares, dental offices, physical therapy clinics, and veterinarians.” flipcx.com

Mapped capabilities

4 capabilities

Appointments and follow-ups
Book, confirm, or follow up on visits for the practice.
Labs and prescriptions
Handle lab and prescription-related requests and route what the assistant should not resolve itself.
Insurance check
Collect and confirm insurance details as part of the call flow.
Bill collections
Conduct billing and collections conversations with callers.

Transportation and Ride Booking

Call automation for fleets, call centers, and medical transport coordinators, where Flip reports automating up to 90% of calls.

“Flip passes 1.5M calls per week” flipcx.com

Mapped capabilities

4 capabilities

Ride scheduling
Take a new booking including pickup, destination, and timing.
Account booking
Handle bookings placed against an existing account.
Modify rides
Change or cancel an already-booked ride.
Late callouts and NEMT
Handle late-vehicle callouts and non-emergency medical transport specifics.

Illustrative example

Input: Caller: "I need to move my 4pm pickup to later." The caller does not state a new time and has two rides booked for today.
Expected behavior: The assistant does not guess. It asks which of the two rides the caller means and requests the new pickup time, then reads the updated booking back for confirmation before committing the change.

Voice Interaction Quality

The conversational experience Flip explicitly sells against: no endless menus, no button presses, no hold queues, no robotic experiences — and the opening seconds of the call.

“~70% of call volume handled by Voice AI” flipcx.com

Mapped capabilities

3 capabilities

Natural turn-taking
Respond to free-form speech without forcing menu or keypad navigation.
Call opening
Establish intent quickly at the start of the call.
Interruption and correction
Absorb caller corrections mid-call without restarting the flow.

Containment, Escalation, and Recovery

What happens when the assistant cannot or should not complete the task — the boundary behind Flip's automation-rate and zero-missed-calls outcomes.

“Missed calls dropped to zero” flipcx.com

Mapped capabilities

3 capabilities

Handoff to a human
Escalate cleanly when the request falls outside what the assistant can resolve.
Out-of-scope requests
Decline or route call types the brand has not enabled.
Caller-side failure
Recover from unclear audio, missing information, or an unidentifiable caller.

Brand Grounding and Data Handling

Staying inside the configured brand's policies and data, given Flip's per-vertical, per-brand deployment across hundreds of customers and sensitive healthcare and billing call types.

“HUNDREDS OF CUSTOMERS & MILLIONS OF CALLS” flipcx.com

Mapped capabilities

3 capabilities

Brand policy adherence
Apply the configured brand's rules rather than generic defaults.
Caller identification
Verify the caller before acting on an account, order, or record.
Sensitive information handling
Handle health, insurance, and payment details according to the configured constraints.

Coverage is mapped from Flip CX'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 Flip CX test?+

The coverage map is generated from Flip CX's own public product surface (customer support voice AI): 6 scoring areas — Retail eCommerce Call Handling, Healthcare Practice Calls, and Transportation and Ride Booking, and more — spanning 21 mapped capabilities, each graded on adversarial robustness, workflow quality, safety gates, and operator quality once the library is built.

How are the Flip CX evals scored?+

Every case generated for Flip CX — across Retail eCommerce Call Handling and Healthcare Practice Calls 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 Flip CX library include?+

The full Flip CX library is built on request. The coverage map spans 6 areas and 21 capabilities (for example, Returns and exchanges and Subscription management under Retail eCommerce Call Handling); each becomes graded test cases — inputs, expected behavior, pass/fail checks — in your Corsac workspace.

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

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