Murf
Also known as: Murf AI, murf.ai, Murf Falcon, Murf voice agents, Murf API, Murf Studio
Murf sells voice AI agents that answer and place calls for customer service and sales, built on Falcon, its own real time speech model. Agents book appointments, update CRM records, take payments and hand calls to people with full context, in more than 35 languages and under 800 milliseconds end to end, on the language model the customer chooses. The Murf API sells Falcon speech at $0.01 per 1,000 characters, alongside speech to text, voice cloning and dubbing.
Murf began as a text to speech studio for voiceovers and now sells voice AI agents for customer service and sales alongside the speech models that power them. More than 10 million users and over 300 of the Forbes 2000 use its products in more than 190 countries, with Honeywell, Oracle and Baker Hughes among the logos, and more than 1,000 teams across customer support, sales, financial services, healthcare, retail and real estate run its agents. The agents run on Falcon, Murf's own real time speech model, which answers in about 55 milliseconds of model latency, holds steady at 10,000 concurrent calls and costs $0.01 per 1,000 characters through the API.
The voice agents answer inbound calls, place outbound calls, qualify leads, book appointments, send reminders, collect information, update CRM records, create tickets and route complex calls to the right person, responding end to end in under 800 milliseconds across more than 35 languages and switching language mid sentence. Murf's agent designers and forward deployed engineers set each agent up around the customer's own procedures, policies and call flows, with a custom prompt, persona, tone and guardrails, and retrieval grounds answers in approved documents, FAQs, policies and help center content.
Mid call function calling reaches Salesforce, HubSpot, Zoho and Pipedrive, Google Calendar, Outlook and Calendly, and Zapier, Make and n8n, over telephony from Twilio, Vonage, Telnyx, Genesys, Exotel or any SIP trunk. Routing rules on intent, sentiment, urgency or account tier hand a call to a person with the transcript and recording attached. The same agents work in website widgets, SMS, WhatsApp and email, run on OpenAI, Anthropic, Gemini or a fine tuned model, and listen through Finch, Murf's streaming speech to text. Agent customers have seen a 40% lower cost to serve and a 30% rise in CSAT.
For teams building their own agents, the Murf API offers Falcon and the more expressive Gen2 voices, streaming speech to text, voice cloning, voice changing, translation and dubbing, with streaming and WebSocket endpoints, SDKs in Python, Node and Go, an MCP server, and plugins for LiveKit, Pipecat and Agora. Falcon can also run on premises through a license agent or as an Amazon SageMaker endpoint.
The company holds SOC 2 Type II and ISO 27001 and 42001 certification, with HIPAA and GDPR compliance. Access is role based with two factor sign in, data is encrypted with 256 bit AES under managed keys, and the core cloud runs in Amazon's Ohio region. Falcon costs $0.01 and Gen2 $0.03 per 1,000 characters, speech to text $0.12 an hour and voice changing $0.10 a minute, while agent deployments are quoted and Studio plans start at $19 a month billed yearly.
The agents are aimed at contact centers, lenders, healthcare providers, real estate firms and sales teams that want a phone agent designed and tuned for them, while the API serves developers who need fast, low cost speech inside agents they build on LiveKit or Pipecat. There is no self serve agent builder or per minute agent price; agents come through a sales conversation.
Vendor details
Canonical URL
https://murf.ai
Category
Voice agent
Subcategory
Voice AI agents for inbound and outbound calls in customer service and sales, built on Murf's own Falcon real time speech model, with a public speech API for teams building their own agents
Funding status
Privately held voice AI company.
Company status
independent
Use cases & customers
Primary use cases
Target customers
Deployment options
Integrations
Salesforce, HubSpot, Zoho and Pipedrive; Google Calendar, Outlook and Calendly; Zapier, Make and n8n; Twilio, Vonage, Telnyx, Genesys, Exotel and SIP trunks; website widgets over WebRTC, SMS, WhatsApp and email; OpenAI, Anthropic and Gemini models; LiveKit, Pipecat and Agora plugins for Falcon; Canva, PowerPoint, Google Slides and Adobe Captivate for Studio; MCP server.
In practice
A dental group puts a Murf receptionist on its main line. The agent answers every call, checks openings in Google Calendar, books the cleaning, sends a reminder by SMS and transfers anyone describing pain to the front desk with the transcript attached.
A consumer lender has Murf call back every applicant who leaves a form half finished, in English or Hindi as the caller prefers, collect the missing details, update Salesforce and route high value applicants to a loan officer.
A startup building its own support agent on LiveKit swaps in Falcon for text to speech, cutting time to first audio below 100 milliseconds and paying $0.01 per 1,000 characters at production volume.
Sources & related URLs
Related / legacy domains
Agentic Index coverage score
10.5 / 14 capabilities · 75%
| Integrations & Tool Calling | Full |
|---|---|
|
Mid call function calling pulls customer data, updates records, books appointments, takes payments and transfers calls through connectors to Salesforce, HubSpot, Zoho and Pipedrive, Google Calendar, Outlook and Calendly, and Zapier, Make and n8n. Calls arrive over Twilio, Vonage, Telnyx, Genesys, Exotel or any SIP trunk without changing numbers, and anything with an endpoint can be called through the open REST API. Sourcemurf.ai/ai-voice-agentread 2026-11-19 |
|
| Workflow Orchestration | Partial |
|
Each agent runs a conversation flow designed around the customer's own procedures, with routing rules on intent, sentiment, urgency or account tier and tool calls chained inside the call, and one orchestration layer spans voice, chat, SMS and WhatsApp. The flows are built and tuned by the vendor's agent designers rather than in a customer facing builder, and no multi agent handoff or branching workflow editor is offered. Sourcemurf.ai/ai-voice-agentread 2026-11-19 |
|
| Knowledge Grounding & RAG | Full |
|
Custom retrieval grounds each agent's answers in the company's approved documents, FAQs, policies and help center content, so the agent answers from the business's own material rather than general knowledge. During a call it also looks up live customer data through function calls into the CRM or other systems, and the prompt and persona carry the company's tone and rules. Sourcemurf.ai/ai-voice-agentread 2026-11-19 |
|
| Human Oversight & Guardrails | Full |
|
Routing rules on intent, sentiment, urgency or account tier send a call to a human agent whenever the conversation needs one, and the handoff carries the transcript and recording so the caller does not repeat anything. Each agent runs with its own guardrails set in its prompt and persona, and retrieval limited to approved content keeps answers inside company policy. Sourcemurf.ai/ai-voice-agentread 2026-11-19 |
|
| Security, Identity & Governance | Full |
|
Murf holds SOC 2 Type II, audited without exceptions, and ISO 27001 and 42001 certification, with HIPAA and GDPR compliance for regulated callers. Access is role based with least privilege, quarterly reviews and two factor sign in, data at rest is encrypted with 256 bit AES under AWS KMS and in transit with TLS 1.2 or later, and security events are kept in a tamper evident log. Sourcemurf.ai/securityread 2026-11-19 |
|
| Observability & Auditability | Full |
|
Every call leaves a log with its recording, transcript, summary, outcome, extracted structured data and latency figures, so a supervisor can trace what the agent said and did, and dashboards can be shaped around the customer's own evaluation framework. Enterprise accounts add access logs, and critical security events are recorded in tamper evident logs that are watched continuously. Sourcemurf.ai/ai-voice-agentread 2026-11-19 |
|
| Memory & State Persistence | Not documented |
|
An agent holds the thread of a call, handling corrections and barge ins without losing context, and passes the full transcript to a person on transfer. Nothing carries from one call to the next on its own: a returning caller is recognized only through whatever the CRM lookup returns, with no agent memory of earlier conversations, preferences or outcomes kept between calls. Sourcemurf.ai/ai-voice-agent/ai-receptionistread 2026-11-19 |
|
| Deployment & Data Residency | Full |
|
Falcon can run on premises, through a local license agent and a TTS server on the customer's own machines or as an Amazon SageMaker endpoint from AWS Marketplace, and region specific processing is offered for Falcon speech. The agent platform is sold for cloud or on premises deployment with data residency across more than ten regions, while the standard cloud stores customer data in the United States, in AWS US East (Ohio). Sourcemurf.ai/api/docs/llms.txtread 2026-11-19 |
|
| Prebuilt Agents / Templates / Packs | Partial |
|
Solutions are packaged by job and industry: AI receptionist, call center, outbound calling, cold calling, AI SDR, recruiter, consumer lending and customer service, with versions for healthcare, real estate and banking. Each agent is still designed around the customer's own procedures and call flows rather than started from a ready template, so the packages name use cases more than agents a team switches on by itself. Sourcemurf.ai/ai-voice-agentread 2026-11-19 |
|
| Triggers & Channel Coverage | Full |
|
Agents answer inbound calls the moment they ring and place outbound calls started through the API, such as calling back a prospect who just filled in a demo form, at up to 10,000 concurrent calls. The same agents sit in website voice widgets over WebRTC, inside an existing chatbot, and on SMS, WhatsApp and email, so a conversation can begin on whichever channel the customer uses. Sourcemurf.ai/ai-voice-agentread 2026-11-19 |
|
| Model Flexibility & Routing | Full |
|
Agents run on the language model the customer chooses, from OpenAI, Anthropic or Gemini or a fine tuned model, and the provider can be switched later. The voice side offers Falcon for real time calls and Gen2 for expressive speech, with more than 150 voices and zero shot cloning for a brand voice, while Finch, the company's own streaming speech to text model, handles listening. Sourcemurf.ai/ai-voice-agentread 2026-11-19 |
|
| APIs / SDKs / MCP Extensibility | Full |
|
The Murf API exposes text to speech over HTTP streaming and WebSocket, streaming speech to text, voice cloning, voice changing, translation and dubbing, described in OpenAPI and AsyncAPI specs with SDKs for Python, Node and Go. An MCP server brings speech generation into Claude and other clients, webhooks report finished jobs, and LiveKit, Pipecat and Agora plugins put Falcon inside agents a team builds itself. Sourcemurf.ai/api/docs/llms.txtread 2026-11-19 |
|
| Testing, Debugging & Optimization | Partial |
|
An evaluation harness supports testing, simulated calls, call data analysis and feedback loops, and the vendor's forward deployed engineers use it to keep improving each agent after launch. Customers see the results through call logs and dashboards, but no self serve test suite, scored scenario runs or side by side comparison of agent versions is offered to them directly. Sourcemurf.ai/ai-voice-agentread 2026-11-19 |
|
| Browser / Computer-use | Not documented |
|
Murf agents act through function calls, connectors, telephony and APIs, so every action they take, from a booking to a payment, runs through a defined integration. No agent drives a browser or a desktop to work through a web portal, a legacy screen or an application that lacks an API, so a task living only behind a screen stays with the staff who handle it today. Sourcemurf.ai/ai-voice-agent/integrationsread 2026-11-19 |
|
The Agentic Index coverage score grades every vendor Full, Partial or Not documented against the same 14 buyer facing capabilities, from public evidence only. Each capability links to how all vendors in the index score on it. How this evidence is graded
Pricing
Voice agents quoted; API from $0.01 per 1,000 characters (Falcon); Studio from $19/mo billed yearly
API by characters, hours or minutes used; Studio by plan; voice agents by quote
Included quota
Studio Creator 24 hours a year of voice generation; Business 96 hours a year
What is public
The Murf API prices Falcon speech at $0.01 and Gen2 at $0.03 per 1,000 characters, speech to text at $0.12 an hour, voice changing at $0.10 a minute and translation at $0.02 per 1,000 characters. Studio runs from a free plan to Creator at $19 and Business at $66 a month billed yearly.
Billing mechanics
API usage bills by characters, hours or minutes as used, and Studio plans bill monthly or yearly, with yearly billing about a third cheaper.
Variable cost rationale
API spend follows the characters spoken and hours transcribed, so cost rises with call volume, while agent deployments carry quoted terms.
Additional watchouts
The free Studio plan carries no commercial rights.
Sales call required
Mixed (some tiers require a call)
Free / trial
Free Studio plan with 10 minutes of voice generation and no commercial rights; startup program with 50 million free API characters for three months
Lowest paid plan
Falcon text to speech API at $0.01 per 1,000 characters
Commercial notes
Voice agents are designed and deployed by Murf's team and priced through sales, and Studio Enterprise adds single sign on, an MSA, no training on customer data and custom voice clones as an add on.
Key ambiguities
Whether language model and telephony minutes come bundled into a voice agent quote is not public.
Support SLA / resale
Studio Enterprise adds a customer success manager, account manager and tailored onboarding.
Missing data
Voice agents carry no public price.
Related vendors
- AgentVoice — AgentVoice builds and runs inbound and outbound AI voice agents, and…
- aiola — Voice agents for field teams on Salesforce
- Aircall — Aircall's AI Voice Agents answer and place calls on Aircall's own…
- Airo by Ringover — Airo is Ringover's AI voice agent for inbound calls, outbound…
- Allo — Allo is a business phone system for small teams with an AI…
- Autocalls — Autocalls is a no code platform for AI voice agents that place and…
Alternatives to Murf
Other products that do the work Murf does, and what each one is.