AI Voice Assistant: Talk to Winston in Your Chat App

An AI voice assistant lets you speak instead of type. With Winston the Pug, you record a voice memo in the messaging app you already use, and he transcribes it, understands it, and replies. No new app to install, no wake word to remember.

Winston lives inside WhatsApp and Telegram, with Signal, Discord and Slack on the way. The voice note button is already in those apps, sitting next to where you type. Hold it, talk, let go, and your words become a request Winston can act on.

“The fastest way to capture a thought is to say it out loud before it slips away.”

How does the AI voice assistant work in WhatsApp and Telegram? #

You send a voice memo the same way you would to a friend. Winston transcribes what you said, works out what you want, and answers in the chat.

There is nothing to set up. WhatsApp and Telegram both have a microphone button built into the chat. You press it, speak in plain language, and release. Winston turns the audio into text, reads it the way he would read a typed message, and replies with what you asked for. If you said “remind me to call the plumber at four”, he sets the reminder. If you rambled through three errands, he sorts them into a short list and reads it back so you can confirm.

A few things worth knowing:

• You can talk naturally. You do not need to phrase things like a search query.
• Winston replies in text by default, so you can glance at it later without re-listening.
• Long memos are fine. He keeps the thread of a longer message rather than only catching the first sentence.

To be clear about what this is and is not: Winston answers voice notes you send in chat. He does not pick up live phone calls. The interaction is a memo you record and send, and a reply that lands back in the same conversation.

When is talking better than typing? #

Voice wins whenever your hands or eyes are busy. Walking, driving, cooking, carrying shopping, or holding a baby are all moments when a keyboard is the wrong tool.

Typing is precise but slow, and it demands both hands and your full attention on a screen. Speaking is quicker for most people, and it frees you to keep doing whatever you were doing. Consider the everyday cases:

• Walking to the station. You remember three things you need to buy. Say them into a memo rather than stopping to type.
• Driving. You think of an email you must send when you get in. Dictate it to Winston, and he drafts it ready for you to send later.
• Hands full at home. Dinner is on the hob and a child is on your hip. Ask Winston to check tomorrow’s school run time without putting anything down.
• First thing in the morning. Before you are properly awake, it is easier to mumble your plan for the day than to peck at a keyboard.

In all of these, the win is the same: you capture the thought at the moment you have it, instead of hoping you will remember it later. Captured thoughts get done. Forgotten ones do not.

What can Winston actually do from a voice note? #

Anything he can do from a typed message. A voice memo is just another way in, so it connects to the same tools and the same memory.

Because the audio becomes text Winston reads normally, voice notes plug straight into his everyday jobs. You can dictate a calendar entry, ask what is on this afternoon, draft an email for him to prepare, or add to a list. For email specifically, Winston drafts and you send, so a voice note never fires off a message behind your back. You stay in control of the real-world action.

This pairs especially well with the things covered in our guide to what an AI personal assistant actually does. Voice is the input; the assistant behind it is what makes the input useful. You can see the full set of connected tools and channels on the features page.

How does voice work with Winston’s memory? #

A voice note is remembered the same way a typed one is. What you say feeds the same long-term memory, so the next conversation already has the context.

This is where speaking becomes genuinely powerful. If you tell Winston in a quick memo that your daughter is called Mia and her swimming lesson is on Tuesdays, he holds onto that. Weeks later you can say “what time is Mia’s lesson” and he knows. You are not re-explaining yourself every time. We go deep on how this works, and how it is kept private, in our piece on an AI assistant with memory.

The same memory also lets Winston be helpful before you ask. A dictated “remind me about the dentist next week” can turn into a gentle nudge at the right moment, which is the heart of a proactive AI assistant that prompts without pestering.

Is it private to speak to my assistant? #

Yes. Your voice notes and the memory built from them are treated as your data, encrypted and kept to you, and you can export or delete it at any time.

Speaking can feel more personal than typing, so the privacy posture matters. Winston keeps each person’s memory separate and encrypted, connected accounts are stored with encryption at rest, and there is no lock-in. You can take your data with you or remove it whenever you choose. The full detail of how this is handled lives on the privacy page and in our article on why AI privacy matters.

The short version: a voice memo to Winston is as private as a voice memo to a trusted friend, with the difference that you can wipe the record on demand.

Getting started with voice #

Speaking to Winston needs nothing extra. If he is in your WhatsApp or Telegram, the voice button is already there. The next time your hands are full and a thought arrives, hold the microphone, say it, and let go. You will have it captured before you would have finished typing the first word. See which apps he works in on the channels page.

Ready to start? Get Winston the Pug. Lives in WhatsApp and Telegram, with Signal, Discord and Slack on the way.