Action execution via registered capabilities
Expose JavaScript functions as capabilities so the agent can trigger real app actions such as cart updates, coupon handling, or other typed handlers from natural-language requests.
Voxide
Voxide is a voice AI SDK for websites that connects spoken or typed requests to JavaScript functions, with live state, routing, and React or headless integration.
Voice AI for browser-based web apps
Voxide is a voice AI SDK for websites that lets you connect natural-language input to your app’s real JavaScript functions. Instead of adding a separate chat experience, you register capabilities that the agent can call when users speak or type, and the SDK runs those actions in the browser.
The product is positioned as a capability-first execution layer for web apps. It can read live state, follow routes, and work with React, Next.js, Vue, Svelte, Angular, Solid, Django, Rails, Laravel, WordPress, and plain HTML, with a prebuilt widget available on the React path and a headless client for other stacks.
Features
Live state awareness
Bind current UI state into the agent so it can respond using what the user sees, including route, cart contents, and other live page data.
Routing and navigation control
Handle navigation as part of the voice workflow, including moving users to routes such as billing or other in-app pages when the request maps to a route change.
React widget and headless client options
Use the prebuilt React widget, which includes a chat panel, voice bar, orb, and visualizer, or build your own interface when using the headless client on other stacks.
Confirmation for dangerous actions
Mark destructive actions as dangerous so the SDK can pause and request UI confirmation before proceeding.
Multiple integration paths
Support different implementation paths through a React package, a core import, or a script tag for server-rendered or static sites.
Use cases
E-commerce actions
Add a voice-enabled shopping flow where users can say things like add an item to the cart, and your registered handler updates the cart in your app.
In-app navigation
Let users move through product pages, billing sections, or account areas by speaking a destination, while the SDK handles the route change.
SaaS dashboards and workflows
Give SaaS users a way to trigger app functions from the browser, such as applying a coupon or invoking other registered capabilities inside an authenticated session.
Internal tools
Connect voice input to internal tools or admin pages where the interface is already built and the team wants the agent to operate existing controls instead of duplicating logic.
Non-React or server-rendered sites
Use the headless integration paths on Vue, Django, Rails, or plain HTML sites when you want voice control without adopting the React widget.
Pros and Cons
Pros
- Maps spoken or typed requests directly to app functions instead of limiting the agent to conversation.
- Works in the browser, which lets the agent read live UI state and use the current session context.
- Supports multiple implementation paths, from React with a widget to headless integration in other frameworks or plain HTML.
- Includes a confirmation step for dangerous actions, which helps protect destructive workflows.
- Has a free starting tier and two paid usage options, giving teams a way to test before committing.
Cons
- The non-React paths are headless, so you need to build your own interface instead of getting the widget automatically.
- The source says a drop-in embed for non-React sites is planned, but it is not available yet.
- Some setup details are framework-specific, so teams will need to follow the relevant docs for their stack.
FAQ
Do I need to build my own voice/chat UI?
In React, Voxide ships with a drop-in `<VoxideWidget />` that includes the mic, waveform animations, and chat transcript. On other stacks, the client is headless and you build the interface yourself.
Is my publishable key safe to use in the browser?
The documentation says the publishable key is safe to expose in frontend code, but you must configure a Domain Whitelist in the dashboard. Voxide uses strict CORS checks to reject unauthorized WebSocket connections.
What frameworks and site types does Voxide support?
Voxide is set up by installing the React package, using the core import, or loading a script tag. The framework docs describe support for React/Next.js, Vue, Svelte, Angular, Solid, Django, Rails, Laravel, Go templates, WordPress, Astro, Hugo, and plain HTML.
How does Voxide pricing work?
The pricing page defines a session as one full voice or text conversation with the agent. The free tier includes 5 sessions, then you can choose a monthly plan or pay per session.
Can Voxide actually control my website, or does it only answer questions?
The source describes Voxide as an execution engine rather than a chatbot: it can call registered JavaScript functions, read live state, and navigate the UI. It is intended to make websites voice-controlled without rewriting the app.
Quick Facts
- Category
- Voice AI SDK
- Primary use
- Adding voice control to websites and web apps
- Supported paths
- React/Next.js widget, headless core import, script tag for server-rendered or static sites
- Frameworks mentioned
- React, Next.js, Vue, Svelte, Angular, Solid, Django, Rails, Laravel, WordPress, Astro, Hugo, plain HTML
- Pricing
- Free tier with 5 sessions, plus Pro and pay-as-you-go plans
- Source domain
- voxide.app
Voxide Alternatives
ByteAsk
ByteAsk is a terminal-first AI coding agent for C and C++ that edits repos and verifies changes with compilers, debuggers, sanitizers, and tests.
CreateOS Sandbox
CreateOS Sandbox is an isolated compute environment for running code and agent workloads in Firecracker micro-VMs with private networking and SDK, CLI, or MCP control.
hob
hob is an independent workspace for coding agents, with local control over sessions, terminals, history, routing, and follow-up work.
Manta AI
Manta AI is an autonomous web app testing tool that maps app behavior, catches regressions, and generates tests from a URL, no scripts or selectors needed.
SonOf
SonOf connects to your repo and PM tool, audits your codebase, and turns approved work into shipped tickets with senior engineering review.
Ghost
Ghost is a terminal-based AI assistant for chatting, code generation, and CLI tasks. Includes free models, supports Linux, macOS, Windows, and is open source.