WhatsGen: AI Texting Practice

Practice the hard texts before you have to send them. These personas push back, use slang, reply late, and call you out. Nobody is watching, so get it wrong here first.

These AI personas do not agree with everything you say. That is the point.

WhatsGen is texting practice for the conversations that actually matter. Pick a mode, get matched with a persona who behaves like a real person, and run the conversation before you have to have it. The chat screen works like the messaging apps you already use, so there is nothing to learn.

FOUR PRACTICE MODES

Social skills. The awkward stuff. Confronting the roommate who never cleans. A friend breakup.

Dating and rizz. Openers, flirting, ghosting recovery, the text after a first date. Playful banter only.

Professional. Asking for the raise. Pushing back on a manager. Handling the coworker who keeps taking credit for your work. Networking without the cringe.

Language practice. The same scenarios in Spanish, French or Brazilian Portuguese, plus English dialects like Singlish and London slang. Your persona weaves corrections and cultural notes into the chat instead of grading you.

40+ PERSONAS WHO TEXT LIKE PEOPLE

A blunt best friend. A date who tests you. A manager who does not budge. Each one texts differently: varied message lengths, realistic delays, read receipts, typing indicators, slang.

They disagree with you. They push back. Some reply slowly. Some double-text. None of them open with "Of course!"

DESCRIBE YOUR OWN SITUATION

Not seeing your scenario? Write it in plain words and get matched with a persona built for that exact conversation.

REVIEW THE CONVERSATION

Tap Review this chat for coaching on what landed and what did not. It is optional and asks for your consent first.

PRIVATE BY DESIGN

Chat runs entirely on your iPhone on Llama 3.2 from Meta (Built with Llama). The model ships inside the download, so chat works the moment you open the app, and your chat messages never leave your device. Voice input uses Apple on-device Speech Recognition, so audio is never uploaded. That applies to iPhones with 6 GB of memory or more. Older iPhones cannot run the model, so their replies are generated by our cloud AI provider (DeepInfra). We ask your permission first and you can turn it off in Settings.

That privacy is what makes the practice honest. You can be clumsy, you can be blunt, and you can say the thing you would never rehearse out loud.

WHATSGEN PRO

WhatsGen Pro unlocks the full app: all four modes, 40+ personas, custom scenarios and Conversation Reviews.

• Weekly: $3.99 per week after a 3-day free trial
• Monthly: $9.99 per month after a 7-day free trial
• Yearly: $59.99 per year after a 7-day free trial, a saving of 50% against monthly

Prices are in US dollars and your local price is shown before you confirm. Payment is charged to your Apple ID at confirmation of purchase. A free trial converts to a paid subscription unless it is canceled at least 24 hours before the trial ends, and any unused portion of a free trial is forfeited when you buy a subscription. Subscriptions renew automatically at the same price unless canceled at least 24 hours before the end of the current period. Manage or cancel in your Apple ID account settings.

Terms of Use: https://whatsgen.app/terms Privacy Policy: https://whatsgen.app/privacy

AI DISCLOSURE

Chat is generated by Llama 3.2 from Meta (Built with Llama), and every reply is screened by a bundled on-device safety classifier. Personas are fictional characters, and nothing they say is professional advice.

Two optional features are processed in the cloud by DeepInfra, each behind an explicit in-app consent prompt:

• Conversation Reviews. When you tap Review this chat and give consent, the message log is sent to generate your coaching summary.
• Custom Persona Creation. Only the scenario description you write is sent, never your chat messages.

The WhatsGen app interface is in English. Language Practice covers Spanish, French and Brazilian Portuguese, plus English dialects.