Google Interview Warmup is gone. Here is what to use instead — especially if English isn't your first language.

What happened

Google launched Interview Warmup in 2022 as part of Grow with Google, alongside its career certificates. You picked a field, answered five interview questions by speaking (or typing), and machine learning transcribed your answer and highlighted patterns in it. In April 2026 Google quietly retired it, with a one-line notice about “spring cleaning.” The address now redirects to a general article on interview preparation, and Google's suggestion is to practise with Gemini Live instead. No replacement product was announced.

That left a lot of people — students, career changers, and a very large number of non-native English speakers — without the one free, low-pressure place they had to practise speaking.

What Warmup did well

It is worth being precise about this, because a replacement should keep it.

• It got you talking. No account, no card, one click and a question. For someone who has never heard their own interview voice, that first recording is the single most useful thing that can happen, and Warmup made it painless.
• The transcript was the point. Seeing your words on screen is confronting in a good way — the rambling sentence, the point you never reached. Warmup put the transcript front and centre.
• You could edit the transcript. When the recogniser misheard a word — which happened more to accented speakers — you could fix it and the highlights updated. Small feature, big trust.
• It never graded you. No scores, no red marks. Warmup was deliberately encouraging: it showed you your most-used words and which talking points you covered, and let you draw the conclusion. For anxious first-timers that tone was exactly right.

What it never did

Warmup was a warm-up. It was not a coach, and it was never designed for the people who needed coaching most.

• No job description. The questions came from a fixed bank per field. If you were interviewing for a specific role at a specific hospital or bank, the questions were generic by design.
• No delivery feedback. It did not measure how fast you spoke, count your “um”s, or notice that you said “I think maybe I kind of led” instead of “I led.” Those are the things interviewers actually react to, and they are the things non-native speakers most need to hear.
• No model answer. You saw your words highlighted; you never saw what a stronger version of your own answer would sound like.
• Nothing for non-native speakers. No awareness that Arabic speakers drop articles, that Hindi speakers add “only,” that Turkish speakers leave the verb until the end. The transcript showed the pattern; nothing explained it or offered the rewrite.

What Offerview does

Offerview was built for the gap in that list. You paste the actual job description, it generates eight questions for that job, and you answer each one by voice in the browser — phone or laptop. Each answer is transcribed and coached on two things, scored separately:

• Content: STAR structure, relevance to the question, specificity — and what was missing (usually a number).
• Delivery: words per minute against the 130–160 conversational band, filler words with timestamps, hedging phrases with the confident rewrite, repeated vocabulary, and first-language interference patterns — for example “I am working here since five years” → “I've worked here for five years.”

Then a model answer, built only from what you actually said, and a session report with the three fixes that matter most. Accent is never scored. The two things Warmup got right — the transcript first, and an encouraging tone — are kept on purpose.

Side by side

Other options, honestly

If what you want is exactly what Warmup was — a free place to say an answer out loud and see it written down — Gemini Live or ChatGPT voice mode with a prompt like “interview me for a nursing role, one question at a time” will do that, and they will push back with follow-ups, which Warmup never did. What they will not do is measure your delivery, know your job description without you pasting it in, or recognise a first-language pattern. Several paid mock interview products exist as well; most focus on native speakers and on content.