AI Assistant

The AI Assistant page is the web interface of your personal assistant, GENI. Besides Telegram you can send and receive messages straight from the browser here.

Web chat

Send commands and questions via the input field at the bottom. If you have the Telegram add-on active, messages are synced, you see conversations across both channels.

Start a new session: click "New chat" to begin a new conversation. Earlier conversations are kept in the history.

Choosing a model (Growth & Pro)

Above the input field are model buttons. Pick the model that fits your task:

Auto is available on all plans. Manual model selection requires Growth or Pro. Credits are charged per the model and length of the message (see The credit system).

Side panel, Conversations & Knowledge base

The side panel has two tabs:

Conversations

• Recent sessions shown by date
• Click a session to load its history
• Delete sessions via the trash icon

Knowledge base

• Upload files the assistant uses as context
• Supported formats: PDF, Word (.docx), images (JPG, PNG, WebP)
• The assistant reads the content and uses it in its answer

Knowledge document

Your personal knowledge document (created during the intake) is shown as a collapsed accordion. Click the header to expand and read it.

• Edit: adjust the document via the "Edit" button
• Download: save it as a text file via the "Download" button

The knowledge document is always loaded automatically as context.

Context per project

When you chat inside a project folder, the assistant combines two sources:

1. Global knowledge document, your personal info, way of working and preferences (always present)
2. Project instructions, the specific context of that project (client, tone, rules)

Set project instructions via the gear icon on a project folder in the sidebar. Conversations outside a project only use the global knowledge document.

Result: a conversation in project "De Krent Bakery" automatically uses a formal tone and brand guidelines, without repeating it each time.

Voice Chat, speak and listen back

Click the microphone icon to the right of the input field. GENI listens, transcribes your question automatically and not only replies in text, it also reads the answer aloud in the voice you set.

How it works:

1. Click the mic in the chat, recording starts
2. Ask your question or give a command
3. Click again to stop
4. GENI transcribes -> processes -> streams a text answer -> plays it as speech

Setting a voice: go to Settings -> Assistant -> Voice. dGENIX uses realistic ElevenLabs voices, with separate Dutch and English voices (a sensible default is set per language). There is also a full Voice Call mode (the sound-waves icon) for a hands-free spoken conversation with an animated orb screen.

Voice answers use more credits than text (text-to-speech is charged per characters spoken). Voice messages via Telegram are also processed: GENI transcribes them automatically and replies.

Uploading files for the assistant

Via the paperclip icon in the input field you can attach files to your message:

• Images -> the assistant views and describes them
• PDF/Word -> the assistant reads the content

Camera in the chat

Click the + button to the left of the input field and choose Camera / Selfie to take a photo without leaving the chat. A live preview opens; the centre button takes the photo, the arrow button switches between front and back camera. After taking it you can retake or confirm; the confirmed photo is attached to your next message. Works on desktop (webcam) and mobile. Your browser asks for camera permission on first use.

Personalised suggestions on the empty chat screen

When you open a new conversation you see 6 suggestion cards. Click one to drop the text into your input field, you can still edit before sending. They are chosen based on your plan and your active skills (and surface skills you could still activate). Use the shuffle button for a new set.

Tools button, modes for deeper answers

Next to the + button is a tools icon. Click to pick from 5 modes that change how your answer is generated:

An active mode shows an indigo chip above the input field with an x to switch it off. Modes that need a specific skill (Deep Research, SEO/GEO) show an upgrade or activate prompt if you do not have it yet. Sources the assistant uses appear as clickable links below its answer.