How to use Prompter Assist
The app listens to your conversation and shows an AI hint in a window invisible to the other party. For accurate hints you need two steps: create a scenario (the conversation context) and attach files to it (resume, job description, product info).
Install and sign in
Before the tutorial — download the app.
Prompter Assist is a desktop app for Windows 10 (2004+) and Windows 11. Download the installer in your account, run the .exe, and sign in with your email account.
During installation Windows may show a SmartScreen warning — click "More info → Run anyway". This is normal: we do not have an expensive publisher certificate (€400/year), so that the app stays affordable.
Create a scenario (or open a default one)
A scenario is the context for the AI: who you are, what kind of conversation it is, and which documents to use.
To open it: the "⋯" button at the bottom of the window → "All settings" → the "Scenarios and prompts" section. To turn scenarios on for a conversation, use the "⋯" → "Scenarios" menu. You can enable several at once — their instructions and files will merge.
Open them in Settings, read the instructions, tweak them for yourself if you like — and be sure to attach files (see below).
If you want to create your own scenario — the "Create scenario" button is in the same section; it opens a separate editor screen. In the system prompt, write: who you are, what the conversation is about, how to answer, and in which language.
Attach files (this matters)
Without files the AI answers in templates. With files — to the point and based on your facts.
Open the scenario in the editor (the "⋯" → "Scenarios" → "Edit scenarios and files" menu, or the "Scenarios and prompts" section in Settings) and press "Attach". Supported formats: PDF, DOCX, XLSX, CSV, TXT, MD. The AI reads the text of the documents before every answer, so the more accurate the materials, the more accurate the hints.
- The candidate's resume (PDF or DOCX).
- The job description — text copied into TXT/DOCX, or a screenshot saved as a PDF.
- Optionally: your portfolio, a GitHub export, a presentation about yourself, notes about the hiring company.
- A description of the product or service: key benefits, pricing, plans, case studies, FAQ, sales script.
- Information about the client: the brief, a summary from their website, the message history, the decision-maker's profile.
Choose an AI model and answer mode
Everything is in the "⋯" menu at the bottom of the window.
Pick the model in the "⋯" → "AI" menu (or in Settings → "Answer model"). Available providers:
- YandexGPT, GigaChat — Russian, no VPN, low price.
- OpenAI (GPT), Anthropic (Claude) — more accurate on complex tasks, higher price. Connected through our server, no VPN needed.
Answer mode (the "⋯" → "Answer mode" menu) — the form in which the AI gives its hint:
- Auto — adapts to the question: a coding task gets a programmer-style answer, an objection gets a negotiation-style one.
- Standard — a ready answer you can say out loud.
- Programmer — highlighted code and what to say while you type.
- Speaker — large key points, easy to read from the screen during a talk.
- Negotiation — what was heard, how to respond, and the next step toward the deal.
When to answer (the "⋯" → "Answer" menu):
- Automatically, after a pause — the AI answers on its own when the other person stops talking (after 1.5 sec by default). Works well for interviews.
- On the "Answer" button — the AI stays silent until you press "Answer" at the bottom of the window or
Ctrl + Space. Good for sales and important negotiations where you do not want an accidental answer.
Spending is deducted from your balance as you go. The price per 1,000 tokens is shown in Settings → "Answer model", next to each model.
Turn on speech recognition
The "Listen" button at the bottom of the window, or Ctrl + L.
Choose who to listen to in the "⋯" → "Listen to" menu (you can change it while listening is stopped):
- Other person — listens to your computer's system sound (what the other person says on the call). This is the main source.
- Microphone — listens to your voice. The AI can answer your lines too, if, for example, you want to practice answering out loud.
- Both — the other person and you at the same time. Selected by default.
Press "Listen" — the button turns into "Stop", and a status such as "Listening to both · 02:14" appears on the left of the top bar. Recognized phrases appear in the chat. If the AI answers on its own, the hint comes after a pause; if it answers on the button, press "Answer" or Ctrl + Space. To ask the AI in your own words, use the text input button in the same bar at the bottom.
Invisibility during screen sharing
The "eye" button in the top bar of the window — it appears when you hover the mouse over the top of the window.
Protection is on by default: the window is not visible when sharing your screen in Zoom, Telemost, Teams, Google Meet, OBS Studio. It is also invisible on a Win+Shift+S screenshot.
If the eye is crossed out — protection is off, the window is visible during sharing, and a red badge stays on in the top bar. Turn the button back on before the call starts.
Hotkeys
Control it discreetly during a call.
Balance & payment
Pay-as-you-go: you only pay for what you use.
Your balance is in the "⋯" menu at the bottom of the app window. Charges are made as you go, for each AI answer. There is no subscription.
Top up in your account on this site (the "Top up balance" button) or right in the app (the "⋯" → "Balance" → "Top up" menu). Payment is via YooKassa, with a Russian card. The balance updates automatically after payment.
Tokens are spent not only on the AI answer, but on the entire context it reads before answering: your scenario, the attached documents, and the dialogue history.
Tips and frequent questions
Download the app and sign in with the same email you registered with here.
Loading…