Jumpspeak's public homepage shows a Spanish relationship scenario asking how someone's day went. That is a concrete conversational task: understand the question, describe something that happened and handle a follow-up. A displayed prompt establishes the advertised setting, not the quality of an actual returned correction.
The documented product adds roleplay, tutor help and feedback around such practice. For this scenario, a meaningful correction would need to preserve the event and timing the learner intended while explaining the language change. Merely replacing a sentence would not tell the learner which part to reuse. No input/feedback pair was generated for this review, so it cannot rate that explanation quality or infer progress from testimonials.
Pick the tutor mode for the actual speaking task
Jumpspeak's
mode guide describes two distinct AI Tutor routes. In
Immersion, the exchange stays in the target language. In
Bridge, the learner can write or speak in a native language and the tutor answers in the target language. For a beginner who cannot yet express a detail in Spanish, Bridge can supply a model phrase; a learner checking unaided Spanish responses needs an Immersion attempt as a separate observation.
| Test question | Bridge mode | Immersion mode |
|---|
| Can I express a missing word in my native language? | The documented route permits it. | The exchange is in the target language. |
| Did I produce the target-language sentence unaided? | A modeled reply cannot establish that. | Record the exact prompt, your own response and whether help was needed. |
The
free-versus-paid guide says Free samples initial lessons and a limited number of tutor conversations, while Premium lists more than 1,000 lessons. The separate minute policy still governs standard Premium's included AI uses; the guide says Premium AI removes daily AI limits. Before upgrading, run the same short scenario in an available mode and inspect the correction, its explanation and the account's remaining counted minutes. We did not complete that account test, so this is a purchase and evaluation checklist, not measured fluency or feedback accuracy.