Frequently asked questions.

Qafka is an AI agent SDK for apps and websites — a universal Agentic UX layer. The agent answers from your documents, navigates users to the right screen or page, runs real actions on your server, and learns from every conversation, in text or voice.

Agentic UX is the step after UX. Instead of designing one fixed path every user must learn, you give each user an agent that operates the product on their behalf, in their own language. The interface adapts to the user — not the other way around.

Web and React Native mobile apps are live today, built on a shared native core. Flutter is on the roadmap. Text and voice work out of the box on every platform.

Paste your URL. Qafka crawls the site, discovers every page and navigation path, and reads your content into its knowledge base — a working agent layer in minutes. Then one script tag puts it live. No rebuild, no redesign, no new platform.

About a day. It's a ~200-line SDK install plus one CLI command to scan your screens; then your team configures documents, tools and routing from a dashboard. No ML team required.

Yes. Tools execute on your own server with your own credentials — Qafka only passes the user's intent and arguments, and your endpoint runs the work. PII is auto-masked and Qafka never stores your secrets. Your data is never cross-used and never trains our models.

A chatbot only talks. Qafka's agent also navigates your screens and pages, calls your APIs, completes real tasks, and turns conversations into product insight. It's an operator, not a Q&A box.

Yes. The same tools you define for text also work in voice, including multi-step flows and cards that render mid-conversation. Voice is priced separately as a metered add-on.

Plans start at $99/month with 100 conversations included, scaling to enterprise deployments with 5,000. Overage rates decrease as you move up tiers. Every plan includes the full platform on web and mobile.