A private AI assistant on WhatsApp: how it works
Hermes lives in the app you already use. Here is how a private AI assistant works on WhatsApp, what she remembers, and why the memory stays on a server you own in the EU.
You already have WhatsApp open all day. So instead of teaching you a new app, intakto puts your assistant where you already work. You message Hermes the way you message anyone else, and she answers with the full memory of your work behind her.
Here is what that actually looks like, and what happens to your data underneath.
You text her like a person, she answers like a colleague who knows you
There is no command syntax to learn and no dashboard to open first. You write in plain language: “What did I promise Anna last session?”, “Draft a reply to the supplier about the late shipment”, “Remind me Thursday to follow up with the new lead.” Hermes reads it, checks what she knows about your work, and answers in the same thread.
Because she has lasting memory, the answer is grounded in your real situation, not a blank slate. She knows who Anna is, which supplier you mean, and what the lead asked for last week. The longer you work together, the less context you have to spell out.
The browser and WhatsApp share one memory. Start a thought on your laptop in the morning, pick it up from your phone on WhatsApp at lunch, and she is the same assistant either way. Nothing resets between the two.
What she can do from a chat thread
A few of the everyday jobs people hand to Hermes over WhatsApp:
- Recall client history before a meeting, so you walk in prepared in half a minute.
- Draft replies, messages, and short documents in your voice, then refine them with you.
- Track commitments: who you promised what, and by when, with a nudge before it slips.
- Pull together a morning summary or a Friday recap of what moved and what is still open.
- Capture a decision the moment you make it, so it is in the memory and not just in your head.
She does these because she remembers, not because of a clever one-off prompt. That is the difference between an assistant and a search box.
Where your messages and memory actually live
This is the part worth slowing down on, because it is where most “AI on WhatsApp” tools and intakto part ways.
Your assistant’s memory does not live on intakto’s servers, and it does not live inside WhatsApp. It lives on a server that is yours, hosted in the EU. The encryption keys are in your account. intakto keeps no copies. WhatsApp is the channel you talk through; the memory of your clients, projects, and decisions sits on your own server, under EU rules.
That means switching is clean. Decide to leave and you take the whole server: every login, every access, all the memory. There is nothing held on someone else’s systems that you have to request back.
Want a channel that does not depend on any one messenger?
WhatsApp is on by default because it is where most people already are. If you would rather not route through a single messenger, you can turn on a private channel from your account. Hermes sets up an encrypted channel for you on Matrix, the open messaging standard, and you talk through the Element X app. The keys are yours, and your messages do not pass through anyone else’s servers. It is built into the account today, not a promise for later.
Either way, the principle holds: the channel is how you reach her, and the memory is what makes her useful. The memory stays yours.
The short version
A private AI assistant on WhatsApp means you get an assistant who knows your work, in the app you already live in, with no new tool to learn. The clever part is not the chat window. It is that the memory behind the chat sits on a server you own in the EU, and moves with you if you ever leave.
intakto is one plan, EUR 29 a month, with a 14-day money-back guarantee. You can read more about how the memory works or what is included.