Build and run voice agents
audentic is a managed platform for voice agents on websites. Start with a public page from your website or describe your business and the agent's job, then review the draft, test a conversation, and publish it on your site. audentic handles the voice connection, model access, website widget, and conversation records.
You can launch a useful voice agent without writing code or managing model credentials. A developer can also embed the same published agent in a React app.
What you can do today
- Create a private draft from one public HTTPS page, then review its source quotes, business facts, and notes before publishing.
- Alternatively, type the details in the editor or talk to the builder to create and refine an agent on the same page.
- Choose a voice and conversation style, then test with your microphone before publishing.
- Add the agent to an approved website with a generated HTML snippet or the React component.
- Let visitors ask questions, interrupt, read captions, and end a conversation.
- Optionally enable current web answers with source links and agent-created on-screen steps. The agent can revise choices and forms as the conversation changes. A separate setting allows custom interactive screens.
- Let visitors submit a form to your Inbox when they want to send feedback or a request, and review saved conversations and submissions in your account.
Your agent is a draft until you publish it. You choose the websites where a published agent can appear. Agent editing requires an audentic account; visitors can use a published widget without signing in.
Choose your path
- Start with your website: Follow the quickstart to create a private draft from a website page, review it, test, and publish. You can also start by voice or type the details yourself.
- Improve an existing agent: Read agent configuration and knowledge.
- Run a live agent: Learn how to review conversations and visitor submissions.
- Build with a coding agent: Give ChatGPT, Claude Code, or another coding assistant the coding-agent guide and ask it to add the documented embed or React component to your website. A supported machine API and MCP integration are planned; the current account API uses browser sign-in and is not a public coding-agent interface.
For quick answers about visitor accounts, microphone access, supported websites, messages, and developer tools, read the voice agent FAQs.
Current scope
audentic currently focuses on website voice conversations. When an owner enables on-screen steps, the agent can create and revise an interface inside the audentic widget and use the visitor's response in the conversation. Owners can separately allow custom HTML, CSS, and JavaScript in an isolated widget frame. A visitor can submit an agent-created form to the owner's Inbox only after reviewing it and pressing Send. The agent cannot edit the rest of your website, fill its existing forms, or run external business actions. The roadmap separates what is available from ideas under development.