Build intelligent chatbots powered by your own knowledge, or connect your favorite AI agents to multiple LLMs through one OpenAI-compatible API.
Kavana provides the infrastructure underneath AI-powered products and agents — one account, one wallet, one API.
Build custom AI assistants that answer questions using your documents, website, product information, policies, and business knowledge.
Connect your applications and AI coding agents to multiple models through one OpenAI-compatible API.
OPENAI_BASE_URL=https://api.kavana.pro/v1
OPENAI_API_KEY=your_keyKavana retrieves the right pieces of your own content before the model writes a single word — so answers come from your business, not generic LLM memory.
Instead of configuring and billing every model provider separately, use a single OpenAI-compatible endpoint and API key for everything.
Zero-cost models for prototyping and light workloads — same OpenAI-compatible endpoint, no billing events.
curl https://api.kavana.pro/v1/chat/completions \
-H "Authorization: Bearer $KAVANA_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "your-model",
"messages": [
{ "role": "user", "content": "Hello" }
]
'}Instead of configuring and billing every model provider separately, use a single OpenAI-compatible endpoint and API key for everything.
Kavana supports the API patterns modern agentic tools expect — including function and tool calling — so agents can use Kavana as their model gateway.
Names are trademarks of their respective owners. Kavana is compatible with these tools via the OpenAI API standard — no partnership implied.
Every request is metered. Your wallet is charged based on actual usage, giving you predictable control over AI costs.
Sign up and get your personal workspace with a credit wallet.
Upload documents, add a website, or generate an API key.
Create a chatbot or point your existing agent at the gateway.
Chat, automate, monitor usage, and control your costs.
Every account starts with a personal workspace. Invite your team when you're ready — members, chatbots, knowledge bases, and billing are all scoped to the workspace.
Kavana ships the controls teams actually ask for — scoped keys, isolated workspaces, and real-time visibility into spend.
Kavana is an AI infrastructure platform with two connected capabilities: a chatbot builder that trains AI assistants on your own documents and website, and an LLM gateway that gives agents and developers one OpenAI-compatible API to many models with built-in metering and billing.
When you upload PDFs, DOCX files, or add your website, Kavana splits the content into chunks and indexes it for vector search. Each incoming question retrieves the most relevant chunks, and the model writes its answer grounded in those chunks — with sources you can inspect.
Any tool that speaks the OpenAI API can use Kavana — including agentic coding tools like Claude Code, Cursor, OpenCode, Continue, and aider. Set the base URL to api.kavana.pro/v1 and use your Kavana API key. Function/tool calling is supported.
Kavana uses prepaid credits: you top up your wallet and every request is metered per token at the configured per-model price. A dashboard shows today's usage, monthly cost, trends, and a full transaction ledger. Rate limits and quotas help you control spend.
Yes. The entire product and this website work in English and Persian, with full RTL mirroring and Persian typography. Chatbots can answer in Persian using your own Persian content.
Build your AI assistant. Connect your agents. Use the models you need. Pay only for what you use.