🟢 Private beta · Personal account

Your WhatsApp, with a brain of its own

Cadences for WhatsApp is a local agent that learns how you write, replies for you when needed, and never exposes your chats to a third-party server. Runs 100% on your machine.

🔒 Your messages never leave your machine. Local encryption. No accounts.

Cadences Agent
How's the deploy going?
Pushing the bundle now, give me 10 min.
✨ Draft generated in your voice · DeepSeek · 1.2s
The problem

WhatsApp doesn’t scale with your life

500 messages a day, 12 active groups, mom asking every 2 hours, clients waiting. You in the middle, re-typing the same thing.

How it helps

An agent that thinks like you

🧠

Its own identity

Upload your texts, voice notes or PDFs. The agent extracts your voice, opinions and writing style. Replies sound like you, not ChatGPT.

Smart drafts

Generates a draft when someone writes. You edit and send, or let it reply on its own in trusted groups.

🔒

100% local

Your WhatsApp Web session lives on your device. AI models can be local (Ollama) or cloud (DeepSeek, Gemini). You choose.

🎯

Per group & per person

Casual tone with family, formal with clients, technical with devs. Each chat has its own persona.

🧾

Long-term memory

Remembers past conversations, facts about each contact, decisions agreed upon. RAG over your own history.

🎙️

Voice notes & TTS

Transcribes incoming voice notes. Can reply with synthesized audio in your voice (optional).

Download

Take it home

Unsigned build — personal use. Code signing and auto-update coming soon.

Requirements: 4 GB RAM, 2 GB free disk. Works with your usual WhatsApp Web.

View source on GitHub →

Newsletter

Get release updates

Once a month or less: major changes, new integrated models, privacy improvements. No spam.

FAQ

What everyone asks

Is it free? +

Yes. The current build is free for personal use. A Pro tier with cross-device sync and bundled cloud models is planned, but the local agent will stay free.

Do my messages leave my computer? +

Your messages and your WhatsApp Web session never reach our servers. If you choose cloud models (DeepSeek, Gemini…) only the message text and approved context goes to the provider you pick. With local Ollama, nothing leaves your machine.

Can WhatsApp ban me? +

The agent uses WhatsApp Web like any browser. It doesn’t touch the official API. WhatsApp may detect and ban abusive accounts; please use it with moderation, no spam, as an assistant for your own chats.

Does it work offline? +

You need internet for WhatsApp Web. With local models (Ollama) AI inference is offline. With cloud models you also need internet for the model.

Is there an Android / iPhone version? +

Not for now. Mobile WhatsApp doesn’t allow external assistants. The current version is desktop (Windows / macOS / Linux) and connects to your WhatsApp Web.