Duolingo is facing an existential crisis as Google Translate rolls out features to tutor users—and even handle live translation as a bonus

Powered by its AI model Gemini, Google’s new beta “practice” option will let English-speaking Google Translate users practice their Spanish or French with custom lessons designed to fit each learner’s level of comprehension. Spanish, French, or Portuguese speakers can also bolster their English skills using the app.

By writing out a learning goal, like getting to know a host family better or going on vacation, the app will tailor lessons to the user. Much like the language-learning app Duolingo, users can use the new Google Translate features to practice both listening and speaking, with helpful hints available when necessary, while also checking their daily progress, Google said in a blog post. 

Shares of Duolingo fell 3% following Google’s announcement before rebounding Wednesday as of midday. The language-learning app, which boasts 130 million active users, has faced increased pressure from AI models like ChatGPT and Gemini, which are also capable of helping users learn a new language. Notably, only about 10% of Duolingo’s users are paying customers, CEO Luis von Ahn told the New York Times earlier this month.

After von Ahn sent a memo declaring the company’s new “AI-first” strategy, he later clarified his comments, saying Duolingo was “hiring at the same speed as we were hiring before.” 

Still, von Ahn said he wasn’t worried about large-language models (LLMs) encroaching on the language learning space.

“Just having conversations in French on something like ChatGPT gets pretty boring after a while. It doesn’t keep you there. We keep you on task with all the gamification,” he told the NYT. 

Duolingo did not immediately respond to Fortune’s request for comment. 

Google also introduced AI-powered live translation features this week for on-the-spot conversations. Translations will be read aloud and presented on the screen with a transcript. Google’s advanced voice-and speech-recognition models are able to isolate a user’s voice so the app can be used even in noisy spaces.

In June, at its Worldwide Developers Conference (WWDC), Apple also announced new live-translation features powered by its homegrown Apple Intelligence that will be available for iMessage, FaceTime, and phone calls.