How much does Smith.ai cost?

Free $0 a month for 25 real calls, then $3.00 a call, with 25 test calls and 50 simulated calls a month. Pro $150 a month for 75 calls at $2.00 a call, 150 calls at $1.80 or 300 calls at $1.67, overage $2.50, $2.30 or $2.17 a call. Enterprise from $500 a month for 300 calls at $1.67 or 500 calls at $1.60, 1,000+ custom, overage $2.17 or $2.10, with full service setup, custom AI prompting and custom integrations by Smith.ai's team. Live agent add on $3.00 per call when used. Month to month with 30 days' notice; a 6 month plan offers rollover calls and a lower rate.

Does Smith.ai support MCP and tool calling?

Partial under the Agentic Index verification standard. Smith.ai publishes some public evidence of tool calling and API or SDK extensibility, but narrower than full Model Context Protocol (MCP) support as defined by the Agentic Index taxonomy. Evidence basis: Developers reach the AI Receptionist through a Zapier app with one New Call trigger that fires when a call completes, and through Make, which together send caller details to CRMs, spreadsheets and contact lists; Zapier actions into AI Recep. Graded from public evidence only against the Agentic Index 14 point capability taxonomy. No vendor pays for placement.

Does Smith.ai have SOC 2, SSO and role based access control?

Partial under the Agentic Index verification standard. Smith.ai publishes some public evidence of security and identity controls, but narrower than the full set of SOC 2, single sign on (SSO) and role based access control (RBAC). Evidence basis: Workspace admins invite users and see each user's email, role and status, which gives a basic access model inside an account. There is no published SOC 2, ISO 27001 or HIPAA attestation and no trust center, and the security page is a respon. Graded from public evidence only against the Agentic Index 14 point capability taxonomy. No vendor pays for placement.

Does Smith.ai give you tracing and observability into agent runs?

Partial under the Agentic Index verification standard. Smith.ai publishes some public evidence of logging or observability, but narrower than full tracing and auditability across agent runs. Evidence basis: Every call appears in the dashboard's call history with a structured summary, a recording that can be played or downloaded, and a full transcript, kept for 90 days, and summaries can be routed to the right team members, email or Slack. The. Graded from public evidence only against the Agentic Index 14 point capability taxonomy. No vendor pays for placement.

Back to vendors

Smith.ai

Q: Who is Smith.ai for?

law firms screening and booking new matters, home services contractors using ServiceTitan, Jobber or Housecall Pro, solo and small service businesses that miss calls, multi location businesses and franchises, marketing agencies reselling an AI receptionist

Also known as: smith.ai, Smith.ai AI Receptionist, Smith AI

Visit site

Voice agentindependentVerified 2026-10-18

Entry priceFree for 25 calls a month; Pro from $150 a month (75 calls); Enterprise from $500 a month (300 calls)Full pricing detail

Smith.ai's AI Receptionist answers a small business's calls around the clock, sorts callers into playbooks for new clients, existing clients and sales, runs intake and lead qualification, books into Calendly, Clio or ServiceTitan, texts follow ups and warm transfers to staff, with a 24/7 human agent network on standby. Quality Studio scores every simulated, test and real call against the business's own goals before and after launch.

Smith.ai, Inc. built its business on a staffed virtual receptionist service with more than 500 North America based receptionists, and now sells an AI first receptionist alongside it, reporting more than 25 million calls handled for over 5,000 businesses. The human first and hybrid services are context here; the subject is the AI Receptionist, which a business sets up and runs from its own dashboard.

Calls forwarded from an existing phone system, such as RingCentral, Dialpad, 8x8, Vonage or Google Voice, reach the AI Receptionist, which blocks more than 20 million known spam numbers and sorts each caller into one of four playbooks: potential new clients, existing clients, sales and everyone else.

Each playbook sets the information to collect, location and service based qualification, custom questions, and the call action, such as booking through Calendly or into Clio, creating records in ServiceTitan, warm transferring to a team member who can accept or let the agent take a message, or sending a follow up text with an intake form or retainer link. FAQs and the business profile supply answers, and self healing FAQs suggest new entries from gaps found on real calls.

Quality Studio lets the business define goals and scenarios, run AI against AI simulations and browser test calls, score every call with pass or fail checks and an AI Quality Index, and accept suggested instruction fixes. Callers who ask for a person, or who match chosen caller types, can go to Smith.ai's 24/7 live agent network as a paid add on.

The Free plan costs $0 for 25 real calls a month, then $3.00 a call. Pro starts at $150 a month for 75 calls at $2.00 each, falling to $1.80 at 150 calls and $1.67 at 300, and adds onboarding and live support. Enterprise starts at $500 a month for 300 calls at $1.67, or 500 calls at $1.60, with custom prompting, custom integrations and setup done by Smith.ai's team. Live agent involvement adds $3.00 per call when used, plans run month to month with 30 days' notice, and up to 10% of calls can be removed from a bill as spam. Workspace admins manage users, and recordings and transcripts are kept for 90 days.

Smith.ai fits law firms, home services contractors and other small service businesses that want every call answered, qualified and booked without staff, with a human safety net and a way to prove quality before going live. It fits less well for teams that want to build their own API actions, run outbound AI calls, choose the model, deploy in a chosen region, or rely on a published security certification.

Vendor details

Canonical URL

https://smith.ai

Subcategory

AI first receptionist for small service businesses, with caller type playbooks, booking integrations, scored quality testing and a human agent backup

Funding status

Smith.ai, Inc., founded in 2015 in Los Altos, California by Aaron Lee and Justin Maxwell; about $27M raised.

Company status

independent

Use cases & customers

Primary use cases

answering every inbound call 24/7 and blocking spamqualifying new leads by location, service and custom questionsbooking consultations and service calls into calendars and practice management systemswarm transfers to staff and handoff to a live agent on requesttexting intake forms, retainer links and follow ups after the call

Target customers

law firms screening and booking new mattershome services contractors using ServiceTitan, Jobber or Housecall Prosolo and small service businesses that miss callsmulti location businesses and franchisesmarketing agencies reselling an AI receptionist

Deployment options

Multi tenant Smith.ai cloudCall forwarding from existing phone systems such as RingCentral, Dialpad, 8x8, Vonage, CallRail and Google VoiceAI first with an optional 24/7 human live agent backup

Integrations

Native connections for scheduling and practice management, including Calendly, Clio Calendar, Clio Grow and Clio Manage, and ServiceTitan; follow up SMS from the AI Receptionist; call summaries routed to Slack and email; a Zapier app with a New Call trigger and Make for sending call data to CRMs and spreadsheets; transfer requests answered in Slack or Microsoft Teams; custom integrations built by Smith.ai on Enterprise.

In practice

A personal injury firm forwards its main line, sets the new client playbook to ask about the accident date and injuries, books qualified callers into Clio and texts them an intake form after the call.

An HVAC contractor connects ServiceTitan so new callers become customer records with the job details, while existing customers with an urgent outage are warm transferred to the on call technician.

A solo accountant writes three Quality Studio scenarios for rescheduling, pricing questions and new clients, runs simulations until the quality score stops failing checks, then turns on forwarding for the free 25 calls a month.

Sources & related URLs

Related / legacy domains

https://smith.ai/ai-receptionist https://smith.ai/pricing/ai-receptionist https://smithaivoiceassistant.freshdesk.com/en/support/solutions https://smith.ai/integrations

Research sources

Agentic Index coverage score

6.5 / 14 capabilities · 46%

Integrations & Tool Calling	Full
Native connections let the AI Receptionist book consultations and service calls through Calendly and Clio Calendar, log intake to Clio Grow and Clio Manage, and create customer records and jobs in ServiceTitan as the call ends, and it sends follow up texts with intake forms, retainer links or thank you notes. Teams cannot define their own tools: custom integrations and live lookups in a case management system are built by Smith.ai's team on the Enterprise plan, and the Zapier and Make apps only pass completed call data outward. Sourcesmith.ai/integration/clioread 2026-10-18
Workflow Orchestration	Partial
Each call is sorted into one of four caller type playbooks, potential new clients, existing clients, sales and all other callers, and each playbook runs its own steps: required and optional details, location and service based qualification, custom questions, then a call action such as booking, transfer, message or follow up text. Advanced branching that adapts mid conversation is configured for customers by Smith.ai's team on Enterprise, and there is no workflow builder or handoff between agents for a team to set up itself. Sourcesmithaivoiceassistant.freshdesk.com/support/solutions/articles/151000199808-how-to-configure-caller-typesread 2026-10-18
Knowledge Grounding & RAG	Partial
Answers come from a business profile of name, website, address and opening hours plus a list of question and answer FAQs entered in the dashboard. Self healing FAQs spot questions the agent could not answer on real calls and suggest new entries in the dashboard and a daily email, which the business approves, edits or dismisses. There is no document upload, website crawl or searchable knowledge store beyond these entries. Sourcesmithaivoiceassistant.freshdesk.com/support/solutions/articles/151000199592-how-to-set-up-and-manage-frequently-asked-questions-faqs-read 2026-10-18
Human Oversight & Guardrails	Partial
On a warm transfer the agent holds the caller, briefs the team member, and lets them accept the call or have the agent take a message, while a cold transfer hands over directly; transfer requests can also be accepted or declined in Slack or Microsoft Teams. Smith.ai's 24/7 live agents can take over when a caller asks for a person or matches a chosen caller type, for $3.00 a call. Self healing FAQ suggestions wait for the business to approve them, but no agent action waits for sign off before it runs. Sourcesmithaivoiceassistant.freshdesk.com/support/solutions/articles/151000228642-configuring-transfer-connection-methodread 2026-10-18
Security, Identity & Governance	Partial
Workspace admins invite users and see each user's email, role and status, which gives a basic access model inside an account. There is no published SOC 2, ISO 27001 or HIPAA attestation and no trust center, and the security page is a responsible disclosure program for researchers. Single sign on and a written list of controls for the AI Receptionist are not offered publicly. Sourcesmith.ai/responsible-disclosure-programread 2026-10-18
Observability & Auditability	Partial
Every call appears in the dashboard's call history with a structured summary, a recording that can be played or downloaded, and a full transcript, kept for 90 days, and summaries can be routed to the right team members, email or Slack. The dashboard tracks call volume, appointments scheduled and quality scores over time. Bookings and records created in other systems are not traced step by step, and there is no audit log of setting changes. Sourcesmithaivoiceassistant.freshdesk.com/support/solutions/articles/151000200039-call-recording-transcriptionread 2026-10-18
Memory & State Persistence	Partial
During a call the agent holds what the caller has said, classifies the caller type, and passes a summary on a warm transfer. Details it collects are written to the call summary, the dashboard call history and connected systems such as Clio or ServiceTitan, where the business keeps its records. No memory the agent reads back on a caller's next call is described, so each call starts from the playbook and the caller's answers. Sourcesmithaivoiceassistant.freshdesk.com/support/solutions/articles/151000199808-how-to-configure-caller-typesread 2026-10-18
Deployment & Data Residency	Not documented
The AI Receptionist runs as one hosted service that businesses reach by forwarding their calls, with no region selection, private cloud or on premise option and no statement of where call recordings and transcripts are stored. The service is aimed at US businesses, its human backup comes from North America based receptionists, and multi location support on Enterprise synchronizes settings across sites rather than separating where data lives. Sourcesmith.ai/pricing/ai-receptionistread 2026-10-18
Prebuilt Agents / Templates / Packs	Partial
Every account starts from the same four caller type playbooks, with required fields already set and slots for optional information, qualification rules, custom questions and call actions that the business fills in for its own services. Quality Studio offers goal and scenario templates for testing, and industry pages for law firms, home services and IT providers show typical setups, but there is no library of finished agents to adopt. Sourcesmithaivoiceassistant.freshdesk.com/support/solutions/articles/151000199808-how-to-configure-caller-typesread 2026-10-18
Triggers & Channel Coverage	Partial
Work reaches the agent as inbound phone calls, forwarded around the clock from the business's existing line on RingCentral, Dialpad, 8x8, Vonage, CallRail, Grasshopper, Phone.com, Talkroute or Google Voice, with more than 20 million known spam numbers blocked first. The agent answers in English and Spanish. It does not place outbound calls, and no webhook, schedule or API request starts a conversation; the Zapier and Make apps fire only after a call has finished. Sourcesmithaivoiceassistant.freshdesk.com/support/solutions/articles/151000201929-how-to-connect-ai-receptionist-to-zapierread 2026-10-18
Model Flexibility & Routing	Not documented
The language and speech models behind the AI Receptionist are not named and cannot be changed by the business; Enterprise customers get prompting written by Smith.ai's own team rather than model choice. What a business can choose is the voice, from a curated selection of natural AI voices, and whether calls are answered in English or Spanish, while tone and handling follow the instructions written into each caller type playbook. Sourcesmith.ai/ai-receptionistread 2026-10-18
APIs / SDKs / MCP Extensibility	Partial
Developers reach the AI Receptionist through a Zapier app with one New Call trigger that fires when a call completes, and through Make, which together send caller details to CRMs, spreadsheets and contact lists; Zapier actions into AI Receptionist accounts are not available. A call request API exists for the separate human outreach campaigns product, and the AI Receptionist has no REST API, SDK or MCP server of its own. Sourcesmithaivoiceassistant.freshdesk.com/support/solutions/articles/151000201929-how-to-connect-ai-receptionist-to-zapierread 2026-10-18
Testing, Debugging & Optimization	Full
Quality Studio, included on every plan, turns goals and scenarios the business writes into pass or fail checks, then scores simulated AI against AI calls, browser test calls and every real call once live. An AI Quality Index averages goal attainment with five fixed caller experience checks and trends over time, and each failing check comes with a suggested instruction change to accept and retest. Free accounts get 50 simulated and 25 test calls a month, rising with plan. Sourcesmithaivoiceassistant.freshdesk.com/support/solutions/articles/151000232639-getting-started-with-quality-studioread 2026-10-18
Browser / Computer-use	Not documented
The AI Receptionist works by phone and text message and reaches calendars and practice systems through native connections; nothing in it operates a browser or desktop for the agent. Browser use is limited to people making test calls from the dashboard, and the website address in the business profile is a fact the agent can share, not a page it navigates. Sourcesmith.ai/ai-receptionistread 2026-10-18

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

Free for 25 calls a month; Pro from $150 a month (75 calls); Enterprise from $500 a month (300 calls)

Monthly plan priced by included calls, with a per call rate that falls as volume rises and per call overage; live agent help billed per call when used

Free tier

Included quota

Free 25 real calls a month; Pro 75, 150 or 300; Enterprise 300, 500 or 1,000 and up, each with matching test calls and twice as many simulated calls.

What is public

Every plan's monthly price, included calls, per call and overage rates, test and simulated call allowances, and the live agent add on rate.

Billing mechanics

Self serve monthly plans billed by call volume, with Enterprise and longer plans through sales.

Cost watchouts

Live agent handoffs cost $3.00 a call on top of the plan; test calls must be marked to avoid billing; custom integrations and custom prompting require Enterprise.

Variable cost rationale

Billing is per answered call rather than per minute, so long calls cost no more, but every call counts including test calls not marked as tests, and live agent handoffs add $3.00 each.

Additional watchouts

Recordings and transcripts are kept for 90 days, so long term records need downloading or a connected system.

Overage / add-ons

Calls past the allowance cost $3.00 on Free, $2.50, $2.30 or $2.17 on Pro, and $2.17 or $2.10 on Enterprise; overages can be switched off, and up to 10% of calls a cycle can be removed as spam.

Sales call required

Mixed (some tiers require a call)

Free / trial

Free plan with 25 real calls, 25 test calls and 50 simulated calls a month, card required at signup

Lowest paid plan

Pro at $150 a month for 75 calls at $2.00 a call, with onboarding and live support

Commercial notes

The Free plan is ongoing rather than a trial, and Enterprise includes a team that builds and runs the setup.

Key ambiguities

The 6 month plan's rollover terms and rate, and pricing above 1,000 calls a month, are quoted through sales.

Cancellation / refund

Month to month with no long term contract; cancellation needs 30 days' notice.

Support SLA / resale

Pro adds dedicated onboarding, an account manager and live phone, email and chat support; Enterprise adds priority support, a success manager and full service setup; reseller, referral and wholesale partner programs exist.

Missing data

Rates above 1,000 calls a month and the 6 month plan's terms.

Agentic Index verified 2026-10-18

Official pricing page Docs

Data confidence: medium