Persona Designer
Define tone, personality traits, and greeting style so every conversation sounds like your brand - warm, professional, concise, or all of the above.
Design an agent's persona, feed it your knowledge, give it tools, pick the voice - then drop it onto your own website with a single line of code. No ML team required.
Name your agent, write its description, pick personality traits and a greeting style - all without code.
Upload documents, connect URLs, wire up MCP servers and HTTP tools so your agent can answer and act.
Choose the LLM, our Lyric or Fantom speech models, and the TTS voice that powers your agent.
Set the widget title, brand colors, avatar, and display settings - previewed live as you edit.
Drop one script tag on your site, or call the agent-session API. The widget appears automatically.
Persona, knowledge, tools, models, appearance, and forms - tuned in one place, with a live preview of your widget the whole way.
Define tone, personality traits, and greeting style so every conversation sounds like your brand - warm, professional, concise, or all of the above.
Upload PDFs and link URLs your agent can reference in real time. Documents are chunked and retrieved automatically during the conversation.
Connect Model Context Protocol servers, HTTP endpoints, and client-event tools so your agent can take action, not just talk.
Pick the LLM that powers intelligence, our Lyric (90+ languages) or Fantom (1600+ languages, low latency) STT, and a TTS voice from the library.
Customize the display name, subtitle, button text, logo, and watermark. The widget previews live as you edit.
Build forms your agent surfaces mid-conversation - book a demo, capture leads, or run an intake flow, posted to your own endpoint.
Configure each agent's information, then wire up call-transfer rules so a conversation can move - with full context - to the right specialist, with routing and escalation handled for you.
Decide when and how a conversation is escalated, then preview the hand-off with a transfer animation.
One agent routes a live conversation out to the right specialist.
Send each conversation to the agent whose role and language match the visitor's intent.
Transcript, detected language, and collected fields travel with the conversation - no repeats.
Define a fallback order so unresolved calls climb to senior agents or a live human.
Pick a template or build your own — your agent surfaces the right form mid-call, fields pre-filled from the conversation.
Click any template to preview it instantly. Your agent picks the right one automatically mid-call.
Forms that fill themselves. Your agent detects intent mid-call, surfaces the matching form, and pre-fills fields visitors just confirm.
Embed the hosted widget with one snippet, or build a fully custom experience on top of the agent-session API.
We host and deploy the agent for you. Paste a single script snippet into any page and the floating voice widget appears automatically.
<script
src="https://api.oshara.ai/widget.js"
data-agent="your-agent-id">
</script>Build a fully custom frontend. Mint a session with your API key and connect your own UI to the live voice agent.
curl -X POST https://api.oshara.ai/api/agents/agent-session/ \
-H "x-api-key: sk_..." \
-H "Content-Type: application/json" \
-d '{ "agent": "your-agent-id" }'Restrict embedding to your allowed origins, expose a language switcher (English, Nepali, and more), and keep every conversation on infrastructure you control.
A voice agent that knows your product, answers in your customer's language, and books a demo right in the conversation.
Qualify visitors, surface a multi-step form, and route hot leads to your CRM through connected tools.
Ground an agent on your internal docs and MCP tools so staff get accurate, on-brand answers - on infrastructure you own.
Design the persona, feed it your knowledge, and deploy to your site in an afternoon - no ML team required.