VoiceScholr docs
VoiceScholr lets you build a live AI agent that talks with your customer by voice and, when vision is on, sees their camera or shared screen about once a second. You write no code, and your customers need no app or account.
Quick start
- Sign in and describe your agent in one sentence, or pick a template. We write its instructions. Paste your prices, policies or manual as its knowledge.
- Press Talk to test it in your browser, with your camera or a shared screen.
- Turn on Public link in the Agent tab. Then share the link, embed it, or add a Talk button to your site: see Add an AI voice agent to your website.
Guides
- Add an AI voice agent to your website: share link, iframe embed, or a one-line Talk button.
- Add an AI shopping assistant to your Shopify store: the agent knows the page, finds products and opens them for the shopper.
What an agent can do today
- Talk in real-time voice, and speak first with the first message you set.
- See the caller's camera or shared screen, about one frame a second, when vision is on.
- Answer from the knowledge you give it.
- Take a snapshot when the caller agrees. You see it in your dashboard; it is deleted after 7 days.
- Search the web and end the call politely, if you switch those tools on in the Tools tab.
- On a Shopify store: know the page the shopper is on, search your products, open one, and let the shopper add it to the cart with a tap.
Models
Gemini 3.8 Live is included free. With your own API key you can switch an agent to Qwen Omni, OpenAI's realtime model, or xAI Grok Voice (voice only). Keys are encrypted and never shown again.
Limits worth knowing
- Free calls are limited per visitor per day and in length: see pricing. On your own key, calls last up to 10 minutes.
- There are no phone numbers: callers use a browser on their phone or computer.
- When every line is busy, the Talk screen says so and retries on its own for about a minute.
Privacy in short
Calls are not recorded. The transcript is kept in server memory so the caller can see it, and deleted 10 minutes after the call. The live stream goes to the AI model provider, Google Gemini by default, whose own terms apply. Full details: privacy.
Need help?
Email hello@voicescholr.org. Last updated 2026-09-30.