Does Zendesk AI handle a language switch during a conversation?
Zendesk AI failed the language-switch scenario. After the customer switched to English, it kept the policy context but answered in French, including the follow-up reply and the holding line before it.
1 of 1 test case failed
Every test case in this scenario has a result.
The test case
Each test case is judged on its own: Pass, Fail, Not gradable, or Untested. The scenario result above is based on this test case.
Language switch to English mid-conversationFailEvidence
Input 1: “Quelle est la politique de retour ?”
Input 2: “Okay, so if I have Plus membership, I'll have a 45-day return window, right?”
Response 1 is a progress message the tool sent for Input 1; it is not part of the answer.
Response 2 answers Input 1.
Response 3 is a progress message the tool sent for Input 2; it is not part of the answer.
Response 4, the final response, answers Input 2.
Oui, c'est bien cela. Avec l'adhesion Plus, vous disposez de 45 jours a compter de la date de livraison, contre 21 jours pour la periode standard.
Transcribed from Proof 1 · Screen recording at 0:20
Every message the tool sent, in order (4)
- Progress message for input 1
Merci de patienter pendant que je recherche cette information.
- Response to input 1 (message 2 of 2)
Voici la politique de retour de Cedarline, et nous sommes desoles pour toute confusion : retours sous 21 jours (45 jours pour Cedarline Plus), article revendable avec emballage si possible.
- Progress message for input 2
Merci de patienter pendant que je verifie ces informations.
- Final response
Oui, c'est bien cela. Avec l'adhesion Plus, vous disposez de 45 jours a compter de la date de livraison, contre 21 jours pour la periode standard.
| ✗ | Expected: Continues in English without losing the conversation's contextFound: The context stays intact, but the post-switch holding line and answer are both in French. |
The case required the conversation to continue in English, but after the switch both the holding line and the answer were in French, even though the policy context held.
Context carried over. The agent keeps the earlier return-policy context when the customer switches to English.
French hold lines. Before each answer, the agent sends a French holding line, including after the language switch.
Configuration and setup
How this tool was set up for the run and what the test needed in place. Each row is a fact from the run's records; a fact the records do not hold is left out, not guessed.
How this scenario is graded
How we decide Pass, Fail and Not gradable. The same rules apply to every tool tested on this scenario.
Each test case gets one result per tool: Pass, Fail or Not gradable. A test case we haven't run yet shows Untested. There are no partial results.
- Pass: the tool did everything the test expected, and nothing it said contradicts the correct answer.
- Fail: at least one expected behaviour clearly didn't happen; the test case's result says which one and quotes the tool.
- Not gradable: our evidence couldn't settle the outcome (for example a record we needed is missing). It is never counted as a fail, and the test case's result says what's missing.
Where this sits in the benchmark
This page is one cell of a larger study: one tool, one scenario. Only this benchmark's frame appears here.
| Level | Name | Scope |
|---|---|---|
| Benchmark | AI Customer Support Chatbots → | 26 scenarios · 14 tools |
| Capability | Multilingual support → | |
| Scenario | Customer switches languages during the conversation → | |
| Tool | Zendesk AI → |
Global scenario definition → · Global capability definition → · Zendesk AI product page →
History of this result
What has happened to this result since it was first published. Runs and grades are never overwritten: a retest or a re-grade publishes a new result and keeps the earlier one readable.
Act on this result
Nothing filed here edits the run or the grade. A challenge opens a review, and a review can produce a new run or a re-grade — which becomes the current result and leaves this one in the history.
You run the same kind of test against your own setup and get the same behaviour.
Agree →Yours behaves differently. Tell us what you got, with a screenshot if you have one.
Disagree →Something here is wrong — a reference value, a transcription, a grade.
Report an issue →Tell us what changed and we schedule a rerun of the failing test case. The old result stays as history.
Vendor notice →The same record is available as structured data through the AI Demos MCP server, with the counts, the coverage and every per-test-case reason carried as fields.
Verify the proof files
These files support this result. Open a file to inspect the original evidence.
File fingerprints (SHA-256)
A fingerprint identifies the exact file used for this result.