Your questions, answered.

OnSubject.AI is a live assistant for conversations that matter. It listens to your call, meeting or interview. It also knows the documents that you gave it. When you press one key, it writes what you should say next, in about a second. It runs on your Windows PC. An optional companion runs on your phone.

Yes. The live demo runs a copy of the app's main window in your browser, with the same AI. Pick a Scenario, ask a question out loud or type it, and press Answer. When you want it in your real calls, with your own documents, install the app and try it free for 7 days.

It's for anyone whose words carry weight: candidates and interviewers, managers who spend their days in meetings, salespeople, consultants, founders, negotiators, and people who face a Q&A or a debate. If you've ever thought of the perfect answer 5 minutes too late, it's for you.

No. It never joins a call, sends a message or speaks. It's a window on your screen (and your phone) that only you see. You decide what to say, in your own words.

They can't, unless you show it to them. OnSubject.AI doesn't appear in the call. By default, the window stays on your screen but is left out of screen sharing and recordings, so the people on the call don't see it even when you share your whole screen. You can turn this off for a Scenario, under Scenario details in the Scenario editor, if you want to share the window. It applies only while that Scenario is selected. It also isn't on the taskbar. It's an icon in the notification area, and you can pick a plain Windows icon so that it doesn't stand out.

It works with anything that plays sound on your PC: Zoom, Microsoft Teams, Google Meet, Webex, Slack, WhatsApp, phone calls through your PC, and videos in a browser. Your microphone adds your own voice. For meetings in a room, your phone can listen.

It transcribes and answers in dozens of languages. For each Scenario, you choose whether it answers in the language of the conversation, or always in one language. Speech recognition is most accurate in widely spoken languages.

There isn't one today. OnSubject.AI runs on Windows 10 and 11. The phone companion works on iPhone and Android.

OnSubject.AI keeps running in the background. It keeps listening, and your keyboard shortcuts and your phone keep working. To bring the window back, click its icon in the notification area. To quit, choose Exit in that icon's menu.

No. If you can write an email and save a file into a folder, you can build a Scenario. The tutorials come with templates that you can copy.

The first words usually appear about half a second to a second after you press the key. The full answer is there within about 2 seconds. Your Scenario is loaded before you ask. The last sentence is transcribed the moment you press the key. Then the answer appears word by word, as it's written. A slow internet connection or a very large set of documents adds a little time.

They come from your Scenario: its instructions, your documents and the conversations that it remembers. OnSubject.AI also uses the conversation so far, including the answers that it already gave you. It writes what you should say, from your facts.

It's told to answer from your Scenario, and to say so when the information isn't there. You can make that stricter in your own instructions (for example, "never invent anything"). Like any AI, it can sometimes be wrong. You always decide what to say.

Yes. For each Scenario, you can set the style (exact words, a line plus bullet points, bullets only, or a single sentence), the length, the language and the tone. Every answer follows GENERAL.md (instructions for every Scenario) and the instructions in that Scenario's START-HERE.md.

Yes. Each question is sent with the answers that it already gave you in this conversation. So a question like "can you give an example?" builds on what you just said, and the new answer stays consistent with it. If you remove an answer that you didn't use, it no longer counts.

Yes. Every answer ends with 3 questions that might follow next, or that go deeper. Click one and it is added to the transcript as Follow-up, then answered. The same buttons are on your phone. They stay with that answer when you look back at it.

Yes, and each Scenario remembers its own. A conversation is saved as you talk, in the Scenario's Conversations folder. When you pick the Scenario again, the conversation is back on screen. Earlier conversations are read like your documents, so new answers stay consistent with what you already said, decided or promised. Clear transcript starts a new conversation. To forget a conversation, delete its file.

No, and that's on purpose. Listening belongs to one Scenario, so everything that OnSubject.AI hears is saved in the right place. To switch, stop listening, pick the other Scenario, and start again.

Click the line and correct it. Answers always use your version. Add names and jargon to the Scenario, so that they're recognised correctly from the start.

Yes, but only when you ask it to. Answer + screen adds a screenshot of your monitor, and so does Photo › Screenshot on your phone. A part of the screen lets you draw a box around just the question. Otherwise, it doesn't capture anything.

It depends on the interview, and you should respect its rules. Many interviews welcome notes. Some hiring processes forbid outside help, and assessments and exams almost always do. If you're not sure, ask, as Radu did in his story. OnSubject.AI works only from what's true about you. It's just as powerful for preparing and rehearsing as it is live.

In many countries and states, you must tell the participants, or get their consent, before you record or transcribe a conversation. Even where it isn't required, it's the respectful thing to do. Check the rules that apply to you, because you're responsible for following them.

Please don't. Exams and tests are designed to measure what you can do on your own. Using any outside help breaks their rules.

No, and it isn't built to. It suggests questions and organises the evidence, and people make the decisions. Use job-related criteria, apply the same scorecard to every candidate, and tell candidates that the interview is transcribed. Some places regulate the use of AI in hiring, so check the rules that apply to you.

It's stored on your PC. Your Scenarios, documents, the search index, conversations and saved sessions are files on your computer. Your account holds only what's needed for the account itself and for billing.

When you ask for an answer, only what that answer needs is sent to the AI that writes it, over an encrypted connection. That means your instructions, the relevant documents or passages, the recent conversation, earlier answers and conversations, and a screenshot or photo if you included one. Your Scenario is also loaded once when you start listening, so the first answer is fast. With AI transcription, speech is sent to be transcribed. With your PC's speech detection, the audio never leaves your PC. Read the details.

No. We only work with AI providers that don't use your data to train their models.

Yes. Your PC's speech detection turns speech into text locally, and every plan includes it. Answers still need an internet connection, because they're written by an AI in the cloud.

Your conversations and documents are files on your PC, so you delete them like any other file. Each Scenario keeps its conversations in its Conversations folder. Delete a file there to forget that conversation, or delete the folder to forget them all.

No. The companion comes from your PC and opens in your phone's browser. You can add it to your home screen, so that it opens like an app.

No. It connects directly to your PC over your Wi-Fi. The connection is encrypted with a certificate that your PC made. The phone is paired with the secret in the QR code.

Guest networks and some office Wi-Fi networks stop devices from seeing each other. Use a network where they can see each other, or your PC's mobile hotspot. When Windows asks, allow OnSubject.AI on private networks. More phone answers.

Create your account in the app, and you get 7 days of Ultimate with every feature. The trial includes 40 answers (15 of them with a screenshot or photo), 3 hours of AI transcription and unlimited use of your PC's speech detection. You don't need a card, and we don't charge you anything automatically.

Each answer that OnSubject.AI writes for you counts. It can come from the button, a hotkey, a typed question, a screenshot, a part of the screen or a phone photo. An answer that you stop or restart doesn't count. The documents that it opens by itself don't count either.

We tell you when you reach 80%. At 100%, listening, the transcript and saving keep working. Answers start again when your month renews, or right away if you upgrade. If you upgrade, you only pay the difference. When your AI transcription hours run out, your PC's speech detection takes over automatically.

Yes, you can cancel from your account. Your plan stays active until the end of the month that you've paid for. Your files stay on your PC.

For a few interviews or important calls a month, choose Starter ($3.99). For meetings or interviews every week, choose Pro ($7.99). For conversations every day, in meeting rooms with your phone, or with big knowledge bases, choose Ultimate ($9.99). Compare the plans.

You need a PC with Windows 10 or 11 (64-bit), an internet connection, and speakers or a headset for your calls. A microphone is optional. Your PC's speech detection uses the speech tools built into Windows.

No. OnSubject.AI is light while it listens. Your PC's speech detection uses a small amount of processor time. Modern PCs easily handle that and a video call at the same time.

Another app may already use that key combination. Pick another one in Settings › Hotkeys, or untick it there, so that the other app keeps it.

Watch the level bar while the others speak. If it doesn't move, go to Settings › Audio & transcription and choose the output device that your call app uses. If the others' words appear twice, turn off the microphone or use headphones.

Yes. Screenshots use the monitor that you used last, or the one that you choose in the ▾ menu. The menu shows a small picture of each monitor, so you can tell them apart. Your phone shows the same pictures when you ask it for a screenshot. You can draw a part of the screen on any monitor.