AI voice agents that answer and make phone calls
Managed agents or bring your own stack — both on one platform. Live in five minutes.
Start building free · Read the docs
- <5 min to your first call
- Real-time response
- Always-on agents
- 11 Indian languages
One agent, every language.
A booking after hours, an appointment in Hindi, a table in Bengali, a delivery in Telugu. Same agent, in six languages your callers actually speak.
Your first call in under five minutes.
Describe your agent
Tell Rymi the job in plain English and it picks the speech, reasoning, and voice stack for you. Or start from a ready-made template and adjust the parts you care about.
Start calling
Test it from your browser in one click, or connect your phone system for live calls. Queue outbound calls in batches and let the agent work the list.
Get insights
Every call comes back transcribed, summarized, and scored. Track sentiment, add a schema to pull the data points you asked for, and see what is working.
Pricing at a glance
$0.05/min for managed agents · billed in 30-second increments · $1 free call credit every month · no card to start. See full pricing.
Voice agent platform, already built.
The pieces of a working phone agent, already wired together. Use the defaults or swap in your own.
- The right voice for every call — A quick FAQ bot and an agent that works through objections run on the same stack. Match the depth to the call. You pay for what each one uses.
- Grounded in your knowledge — Upload your docs, policies, and FAQs. Your agent answers from what your business knows and replies in real time.
- Speaks your customer's language — Run agents on phone lines, in the browser, or inside your own product. 150+ languages, with natively trained voices for eleven Indian ones.
- Insights after every call — Every call comes back as a transcript, a summary, and a sentiment read. Add a schema and it also extracts the data points you asked for. You never replay a recording.
- Built-in compliance — Do-not-call filtering, recording controls, and per-workspace data isolation. Rymi enforces all three on every outbound call.
- Connected to your tools — Your agents book on your calendar, pull CRM contacts, and send WhatsApp or Telegram. Hundreds run at once, around the clock.
Voice agents built for your workflow.
Inbound pickups, cold calls, reminders, patient follow-ups. These are the calls teams hand to a voice agent instead of staffing a phone all day.
- Front Desk Receptionist
- Appointment Setter
- Lead Qualifier
- Sales Outreach
- Customer Support Agent
- Reminder & Confirmation
- Real Estate Buyer Agent
- Restaurant Order Taker
- Healthcare Receptionist
- Survey & Feedback
- Compliant Payment Collection
- Student & Alumni Outreach
- Delivery & Dispatch Notifications
- Recruitment Phone Screening
- Insurance Claims & Intake
- IT Helpdesk Support
- Hotel & Hospitality Concierge
- Legal Client Intake
Plug in your stack.
Your agents book on your calendar, look up a caller in your CRM, raise a support ticket, and message on WhatsApp or Telegram — mid-call, while the caller waits. Wire each one up once.
- Google Calendar — read availability and book the slot
- Outlook Calendar — book on Microsoft 365
- HubSpot — look up contacts
- Zoho CRM — look up contacts
- LeadSquared — look up leads
- Freshdesk — raise a ticket, or read back its status
- WhatsApp — send confirmations and links
- Telegram — ping your team the moment a call ends
Running something that isn't on this list? Two ways in. Point an agent at an HTTPS webhook — Slack, Teams, Zapier, n8n, or your own API — and it decides when to fire and what to send. Or bring any Model Context Protocol server and its tools join the call, with no integration to build.
Drive Rymi from your AI agent.
Connect Rymi over MCP and your assistant builds agents, queues calls, and reads the results back — in plain English, from Claude, Claude Code, Cursor, ChatGPT, Windsurf, or any MCP client. Point it at https://mcp.rymi.live and pass your Rymi key as a Bearer token, or run @rymi/mcp locally. Read the MCP docs.
Before you start.
How fast can I get started?
Under five minutes. Pick a template, customize your agent, and place a real call from your browser. No phone number or setup required.
How does pricing work?
You pay per minute of talk time, billed in 30-second increments. The free plan includes $1 in monthly call credit, so you can build and test before you commit. You can build agents two ways. A managed stack runs at a flat per-minute rate from $0.05/min, with the speech, model, and voice tuned for you. A custom stack lets you pick every component yourself or bring your own provider keys, for full control over how each agent runs. Mix both across agents.
What's included in the per-minute price?
Your per-minute rate covers the AI conversation, call recordings, transcripts, summaries, sentiment analysis, and analytics. Carrier charges for phone lines are billed separately by your telephony provider. Browser calls cost nothing beyond the per-minute rate.
Do I need special phone equipment?
No hardware required. Test calls run straight from your browser. For live phone calls, Rymi connects to your existing telephony provider, so you keep your own numbers and carrier rates. Nothing to install on your end.
What happens after each call?
Every call produces a full transcript, an AI-written summary, sentiment analysis, and any data you asked the agent to capture. Read it all in your dashboard, or have Rymi push it to your own systems through webhooks the moment the call ends.
Can it speak languages other than English?
Yes. Pick the languages your callers speak in Studio, and Rymi builds a matching speech-to-text, reasoning, and voice stack for each one. Your agent holds the conversation in the caller's language and switches between the ones you enabled.
Can the agent take actions during a call?
Yes. An agent can check availability and book on Google or Outlook Calendar, look up a caller in HubSpot, Zoho CRM, or LeadSquared, raise or check a Freshdesk ticket, and send WhatsApp or Telegram messages. For anything else, point it at your own HTTPS webhook or connect any MCP server. You wire these up once in the agent's configuration.
Is there an API and SDKs?
Yes. Rymi is API-first. Official open-source SDKs cover Node.js (@rymi/node on npm) and Python (rymi on PyPI), plus an MCP server (@rymi/mcp) that lets AI clients like Claude and Cursor build and manage agents directly. Read the developer docs at docs.rymi.live or browse the source at github.com/rymi-live.
Can my customers tell it's AI?
Agents run on realtime voice models that reply in real time, with natural turn-taking and interruption handling instead of long pauses. Responses land fast enough that callers stay in the conversation rather than hanging up on dead air.
Ready to put AI on the phone?
Create your first agent and make a test call in under 5 minutes. From $0.05/min — no card, no contract, nothing to install. Start building free.