Smith.ai
Outsourced front office combining AI receptionists with a live human agent network, positioned by the vendor as AI and live agents working together to handle every call end to end rather than a receptionist bolted onto a chatbot. The company is the dominant name in legal receptionist services, with agents trained on attorney style intake scripts, and also serves medical, contracting and other professional service markets. Products are sold in two lines. AI Receptionist is an AI only tier the vendor describes as deeply context aware and connected to the customer's key systems, including an intake agent handling complex, sensitive and conditional intake, with live agent involvement available as a per call add on on any plan. Virtual Receptionist is the hybrid service where human agents answer calls, greeting callers in the business name, following custom call handling rules, taking messages, booking appointments, transferring urgent calls, qualifying leads against the firm's intake script, and routing enquiries into the CRM, with AI supporting data entry and CRM updates. Coverage is 24/7 across inbound calls, web chat with live agents responding to website chat, outbound calling on higher plans for sales follow up and appointment confirmation, and CRM side intake routing, with emergency keyword routing and free spam and sales call blocking. Service is fully bilingual in English and Spanish. Integrations are stated at more than 7,000, with native connections to Clio, MyCase, LawPay, Salesforce and Calendly plus Zapier, syncing conversations to customer systems in real time, and retainers can be sent automatically once a receptionist completes intake. Pricing is published in detail: an AI Receptionist free tier at $0 per month covering 25 calls then $3.00 per call, AI Receptionist plans from $95 per month, hybrid Virtual Receptionist from $292.50 per month for 30 calls with published per call rates thereafter, and annual price locked Done-for-You AI plans at $500, $1,000 and $2,000 per month with no overage charges and a dedicated success manager. Self serve plans are month to month with no contract, and a 14 day money back period is offered on virtual receptionist service. Named customers with published case studies include Maria Jones Law Firm and SummitPoint LLC.
Capability grades
All 15 axes, graded from public sources on the date shown. Hover a grade to see what the letter means on that axis.
AI Centrality
How much of the product is actually AI. Whether the machine learning is the mechanism the buyer is paying for or a feature layered onto conventional software, and whether the vendor is specific about which is which.
MEMBERSHIP NOTE FIRST, since the brief flagged this name to check the AI bar and described it as substantially a human answering service with AI attached. That description was correct when written and has been overtaken. Smith.ai now sells a distinct AI only product line: an AI Receptionist with a free tier at 25 calls and paid plans from $95 per month, described by the vendor as deeply context aware with an intake agent handling complex, sensitive and conditional intake, and annual Done-for-You AI plans priced separately from the human service. That is a shipped, separately purchasable AI product, not an AI roadmap, so the bar is cleared. The second limb also passes: agents are trained on attorney style intake scripts and the platform integrates natively with Clio, MyCase and LawPay, which are legal specific systems, so this is not horizontal software with no legal product despite the vendor also serving medical and contracting markets. Graded C rather than higher because the company is a human answering service that added AI: independent review describes Smith.ai receptionists as human agents backed by AI tools for data entry and CRM updates, the hybrid remains the flagship for legal intake, and removing the models leaves a working live agent answering service, which is the business the brand is built on.
Citation Accuracy and Hallucination Disclosure
Whether the vendor publishes measured accuracy on citations and assertions, grounds output to primary sources, and says plainly what its system does when it does not know. Legal has a documented public record of fabricated citations reaching filed briefs, so an untested claim of accuracy is not evidence.
Nothing published, and the axis applies in an unusual form. The product generates no legal citations and performs no research, so the classic grounding question does not arise. What does arise is that an AI receptionist answers callers' questions about a firm and their situation in real time and conducts what the vendor calls complex, sensitive and conditional intake, and the accuracy of what it tells a caller is unaddressed. No accuracy figure, no error rate on intake capture, no evaluation, no statement of grounding against firm supplied information, and no hallucination disclosure were located. The failure mode is a caller told something wrong about a firm's practice areas, availability or their own matter, with no artifact recording it. Checked the home page, the AI Receptionist page, the pricing material and independent review on 29 Aug 2026.
Autonomy and Oversight Model
What the system decides on its own, what a lawyer must approve, and whether the vendor documents where the review point sits. A tool that drafts under review and a tool that files without one are different products and different risks.
A hybrid architecture that answers this axis structurally, with a documented autonomy behaviour the vendor does not disclose. On the credit side: live agent involvement is available as a per call add on on any plan, the AI is stated to be seamlessly integrated with the Live Agent Network, and on hybrid plans human agents answer every call, so a human is available at the point of contact by design rather than by policy, which is the right control for a system speaking to distressed callers in high emotion practice areas. THE UNDISCLOSED BEHAVIOUR: independent review reports a consistent pattern of complaints, concentrated on one review platform, that the AI Receptionist auto escalates calls to live agents without user authorisation, driving up bills unexpectedly. That is the AI making an autonomous handoff decision with a direct financial consequence for the customer, reported by third parties and described nowhere in vendor material. Held at C on that basis: the escalation path exists and works, and the rule governing when the machine invokes it is neither published nor apparently controllable by the customer.
Operational and Outcome Evidence
Named, dated evidence that the product works in production at real firms or legal departments. Case studies with figures and identified customers count. Unattributed testimonials and launch announcements do not.
Named customers with named individuals and published case studies, plus the deepest independent evidence base in this category. Vendor published and attributed: Paulina Quiñones of Maria Jones Law Firm quoted on faster response and screening of prospective clients with a linked case study, and Valentin Reyes of SummitPoint LLC quoted on a 60 percent year over year sales profit increase with a linked case study. Independent evidence is unusually extensive and includes material unfavourable to the vendor, which strengthens rather than weakens it: presence across G2, Trustpilot and Clutch with a directory listing on a named legal technology publication, multiple independent comparison analyses, and a third party study of more than 500 billing patterns concluding that real invoices typically run 20 to 30 percent above base plan price. Held at B rather than A because the published outcome figures carry no methodology, baseline or attribution of causation, a 60 percent sales profit increase in particular being a company result rather than a demonstrated product effect, and because no usage or volume figure for the vendor itself was located.
Privilege and Confidentiality Posture
How client confidences are handled: attorney client privilege and work product treatment, segregation of one client matter from another, whether client data trains any model, and what the vendor commits to in writing rather than in marketing.
Nothing located. No confidentiality statement, no encryption reference, no data handling description and no treatment of privilege or prospective client confidences was found. The exposure here is broader than for a chatbot vendor because human agents are in the loop: live receptionists hear callers describe legal problems in real time, including in the high emotion practice areas independent review identifies as this vendor's strength, being personal injury, family law and criminal defence, and nothing published addresses agent confidentiality obligations, training, screening or what an agent may record and retain. Web chat is also answered by live agents. The prospective client confidences question that runs through this category applies with the additional feature that a human third party is the recipient. Checked the home page, the AI Receptionist page, the pricing material, the directory listing and independent review on 29 Aug 2026.
UPL and Professional Responsibility Posture
Whether the vendor is clear that it supplies a tool rather than legal advice, who its audience is, and how it addresses unauthorized practice of law, competence and supervision duties, and jurisdiction limits. ABA Formal Opinion 512 is the reference point.
Not located. Non lawyer agents and an AI conduct intake conversations with prospective clients, qualifying them against a firm's criteria and, in the AI configuration, handling what the vendor describes as complex, sensitive and conditional intake. Nothing published states that agents or the AI do not give legal advice, describes what they may and may not say when a caller asks a substantive question, or addresses the professional responsibility framework a firm operates under when outsourcing first contact with prospective clients to a third party. The gap is notable given the vendor's own positioning as the dominant name in legal receptionist services and its training of agents on attorney style intake scripts, which shows the legal context is understood without the professional rules being engaged. Checked the home page, the AI Receptionist page, the directory listing and independent review on 29 Aug 2026.
AI Governance and Bias Disclosure
Published governance over model behaviour: who owns it inside the vendor, what is tested before release, and what is disclosed about disparate output across matter types, parties, or populations.
Nothing published about how the models are governed, evaluated or monitored. No AI policy, no model card, no bias or fairness testing, no evaluation methodology, no accuracy monitoring, no drift statement, no named governance body, no ISO 42001 and no EU AI Act positioning were located. Two specific risks are untested. The AI performs lead qualification against firm criteria, so it decides which callers are worth the firm's attention, and the same concern recorded on Gideon applies: the people assessed are consumers seeking help who are never told a machine assessed them. Separately, the service is bilingual in English and Spanish and speech systems are known to perform unevenly across accents and languages, and nothing published indicates that recognition or qualification performance has been examined across the caller populations the product serves. Checked the home page, the AI Receptionist page and independent review on 29 Aug 2026.
AI Safety and Data Stewardship
Retention, deletion, access control, and what happens to prompts and documents after they are processed. Whether the vendor states its subprocessors and its incident practice, or leaves the buyer to assume.
No stewardship position located. Nothing states whether call recordings, transcripts, chat conversations or captured intake data are used to train or improve models, no retention period is published, and no deletion right is described. Voice makes this materially weightier than for the text based products in this category: a call is recorded or transcribed to be actioned at all, the vendor's own home page offers playable real call recordings by industry as a demonstration, which establishes that recordings are retained and repurposed for marketing, and nothing published states on what basis, with whose consent, or whether customer calls could be used the same way. Recording consent is itself regulated differently across states. Checked the home page, the AI Receptionist page, the pricing material and independent review on 29 Aug 2026.
AI Liability and Recourse
What the vendor stands behind contractually when its output is wrong. Indemnities, caps, carve outs, insurance, and whether any of it is published or only reachable through a negotiated agreement.
No published position located on liability for AI output, warranty, service levels or remedy, and this is the record where a concrete financial consequence is documented by third parties rather than hypothesised. Independent review reports consistent complaints that the AI auto escalates calls to live agents without authorisation, and a separate analysis of more than 500 billing patterns concludes that real invoices run 20 to 30 percent above base plan price. Together those describe a system whose autonomous decisions increase what the customer pays, with no published rule governing the decision and no stated recourse. Beyond billing, the ordinary exposures apply and are unaddressed: a missed or mishandled call from a prospective client, or a wrongly disqualified caller. A 14 day money back period is published for the virtual receptionist service, which is a return policy rather than a liability position. Checked the home page, the pricing material, the directory listing and independent review on 29 Aug 2026.
Practice Systems Integration Depth
How deeply the product reaches into the systems legal work already lives in: document management such as iManage and NetDocuments, Word and Outlook, contract lifecycle management, matter management, e-billing, and court filing systems.
THE STRONGEST INTEGRATION POSITION IN THIS CATEGORY AND AMONG THE BEST ON THE INDEX. Named legal specific integrations: Clio, MyCase and LawPay, covering practice management, case management and legal payments, plus Salesforce and Calendly, with Zapier extending further and a stated total exceeding 7,000 integrations, corroborated at more than 1,000 by an independent legal technology directory. Conversations are stated to sync to customer systems in real time rather than in batch. The capability that lifts this to A is the completed workflow rather than the connector count: retainers can be sent automatically once a receptionist completes intake, which carries a prospective client from first contact through qualification to engagement paperwork without a human touching the handoff, and independent review confirms the Clio integration as plug and play with no technical setup required. That a competitor's comparison material cites Clio dependent workflow as a reason to choose this vendor is external confirmation the integration works. Held short of perfection because no API documentation was located and the 7,000 figure is unenumerated.
Deployment Model and Data Residency
Where the software runs and where the data sits. Multi tenant cloud, single tenant, private deployment, on premises, and whether region of residence is a published option or an enterprise conversation.
Nothing located. No hosting provider is named, no region or data residency commitment is published, and no deployment options are described. The question is sharper than for a software only vendor because this service holds voice recordings and transcripts of calls from members of the public, and because a live agent network implies human agents located somewhere, which is itself a data location question that nothing published addresses: a firm cannot determine from public material which jurisdiction its callers' conversations are heard and stored in. Checked the home page, the AI Receptionist page, the pricing material and the site navigation on 29 Aug 2026.
Security Certifications and Trust Center
Independent attestation a buyer can pull without a sales call: SOC 2, ISO 27001, penetration test summaries, a trust center with current reports and named scope rather than a badge image.
No certification, attestation, trust centre or security page was located. No SOC 2 of either type, no ISO 27001, no named auditor, no penetration testing partner and no encryption statement were found across the pages read. Under the three tier test the artifact is absent rather than gated. Three records in legal-intake-and-client-development now sit at D on this axis, LawDroid, Gideon and this one, and the pattern is worth watching against the remaining four rather than concluding from three. The gap is material here because the vendor holds call recordings and transcripts on behalf of law firms whose own obligations attach to that material, and because a service handling consumer voice data at this scale would ordinarily be asked for an attestation in procurement. Checked the home page, the AI Receptionist page, the pricing material, the directory listing and the site navigation on 29 Aug 2026.
Model Supply Chain Disclosure
Which models sit underneath, whose they are, where they run, and whether the vendor commits to telling customers when that changes. A legal buyer inherits every dependency it cannot see.
Nothing located. No foundation model provider, speech recognition provider, model family or version is named, and no subprocessor list was found. The gap is broader here than for text based peers because a voice product necessarily involves speech to text, text to speech and language model components, each of which is commonly supplied by a third party, and none is identified, so a firm cannot determine which parties process recordings of its prospective clients' calls. The vendor describes the AI as deeply context aware and connected to key systems, which characterises capability rather than provenance. Checked the home page, the AI Receptionist page, the pricing material and independent review on 29 Aug 2026.
Commercial Transparency
Whether a buyer can learn what this costs without entering a sales process: published rates, the unit being charged, what sits behind an enterprise tier, and what implementation adds.
The most detailed published pricing in the pull, across two product lines and multiple tiers, with the unit of charge stated throughout. Published: an AI Receptionist free tier at $0 per month for 25 calls then $3.00 per call, AI Receptionist plans from $95 per month with per call rates within tier and a stated overage rate, hybrid Virtual Receptionist from $292.50 per month for 30 calls with published per call pricing after the monthly minimum, and annual price locked Done-for-You AI plans at $500, $1,000 and $2,000 per month explicitly with no overage charges. Terms are published too: month to month with no contract on self serve plans, a 14 day money back period, and per call rather than per minute billing, which independent review notes favours firms with longer intake conversations. A prospective buyer can model their own cost from published material. Graded A on that basis. The note records the real qualifier: independent analysis of more than 500 billing patterns concludes invoices typically run 20 to 30 percent above base plan price, so published rates and realised cost diverge. That divergence arises from usage and from the auto escalation behaviour graded on Autonomy rather than from concealed rates, which is why it does not reduce this grade, and a buyer should read the two notes together.
Firm and Practice Coverage
Who the product is actually built for. AmLaw, midlaw, small firm and solo, in house departments, government and courts, and which practice areas are supported rather than merely claimed.
Channel and market coverage are broad and legal practice coverage is not characterised. Channel coverage is the most complete in this category: 24/7 inbound calls, web chat answered by live agents, outbound calling on higher plans for sales follow up and appointment confirmation, and CRM side intake routing, with emergency keyword routing and spam blocking, delivered fully bilingually in English and Spanish. Bilingual delivery is a real access dimension no other record in this category states. Market coverage extends well beyond law into medical, contracting and other professional services, with the vendor's own home page demonstrating an industry scenario outside legal and one independent reviewer observing that the product now prices itself for contractors rather than law firms. Within legal, independent review identifies strength in high emotion intake including personal injury, family law and criminal defence, and firm suitability at 30 to 150 calls per month. Held at C because none of that legal characterisation comes from the vendor, no practice areas or jurisdictions are named in vendor material, and the horizontal breadth means the legal offering is a configuration rather than a described scope.
Legal Signals
What each signal meansA signal records what public sources say on the date shown. It is not a grade and it is not a recommendation. Where a signal reads Not addressed, it means the index did not locate the material in public sources on that date, which is a statement about disclosure rather than about the product.
Client Data in Training
Can material a lawyer puts into this product be used to train a model?
No located term or policy addresses the question either way.
Silent. The quoted line is the vendor's central architectural claim and it describes who handles a call rather than what happens to the recording of it. No statement in either direction was located on whether call recordings, transcripts, chat conversations or captured intake data are used to train or improve models. One published fact makes the silence concrete rather than theoretical: the vendor's own home page offers playable real call recordings by industry as a product demonstration, which establishes that recordings are retained and repurposed beyond handling the call, and nothing states on what basis, with whose consent, or whether a law firm's own calls could be used the same way. The people recorded are members of the public contacting a law firm, who are not the customer and cannot consent, object or ask. Recorded as silent, not as a negative commitment. Checked the home page, the AI Receptionist page, the pricing material, the directory listing and independent review on 29 Aug 2026.
Prompt and Output Retention
How long does the product keep what a lawyer typed, and can that be set to zero?
No located public material states how long prompts and outputs are retained.
Not addressed. No retention period is published for call recordings, transcripts, chat logs or captured intake data, and no deletion right is described. Retention is established as a fact by the product itself, since real call recordings are published on the vendor's home page as demonstrations, and conversations are stated to sync into customer systems in real time, which is transmission rather than a statement about what the vendor keeps. Recording retention carries state law implications on consent and on how long voice data may be held, and nothing published engages any of it. Checked the home page, the AI Receptionist page, the pricing material and independent review on 29 Aug 2026.
Ethical Walls and Matter Segregation
Does retrieval respect the firm’s ethical walls, or can the model read across them?
No located public material addresses walls or matter level segregation.
Not addressed, with a dimension no software only record in this category has. No permission model, access restriction or segregation description was located. Beyond the usual questions, this vendor operates a Live Agent Network, so human agents employed by a third party hear prospective clients describe their legal problems, and nothing published states whether an agent handles calls for competing firms, whether agents are screened or restricted across customers, or what prevents an agent who took an intake call for one party to a dispute from taking one for the other. The prospective client conflicts problem recorded on LawDroid and Gideon applies here as well, and the human agent layer adds a conflict surface that no permission model can address because it lives with people rather than with data. Checked the home page, the AI Receptionist page, the directory listing and independent review on 29 Aug 2026.
Third Party Request and Subpoena Notice
If someone subpoenas the vendor for a firm’s data, does the firm hear about it first?
No located term or policy addresses third party requests for customer data.
Not addressed. No government or law enforcement request clause, no commitment to notify a customer before producing their data, and no transparency report were located. The material held is unusually sensitive: recordings and transcripts of members of the public describing legal problems to what they believe is a law firm, in practice areas independent review identifies as including criminal defence and family law, where a caller's own account could be adverse to their interests if produced. A request served on the vendor would reach voice recordings the firm itself may never have heard. Checked the home page, the AI Receptionist page, the pricing material and the site navigation on 29 Aug 2026.
Primary Law Corpus Provenance
Where does the law in this product come from, and does the vendor have the right to use it?
No located public material identifies the corpus behind the product’s answers.
Not addressed, and inapplicable in the usual sense with a live residue. The product has no primary law corpus: the AI operates on the firm's own intake script, call handling rules and connected systems, so the substantive content is customer supplied. The residue is what the underlying speech and language models were built on, which is entirely undisclosed, and it matters more for a voice product than for a text one because speech recognition performance is corpus dependent and the service is delivered bilingually in English and Spanish to callers of every accent and speech pattern. Nothing published names a training corpus, a speech provider or any evaluation across caller populations. Checked the home page, the AI Receptionist page and independent review on 29 Aug 2026.
Good Law Verification
Does the product tell you when the authority it just cited has been overruled?
No located public material addresses whether authority is checked for subsequent history.
Not addressed, and inapplicable on the facts. Smith.ai answers calls, qualifies callers against a firm's intake script, books appointments and routes enquiries into a CRM, and produces no legal research or citation to authority, so there is nothing for a citator to check. Recorded as a scope fact rather than a disclosure failure, consistent with the treatment on Gideon in this same category and on Tavrn, DigitalOwl, Legal Tracker and Mitratech elsewhere. Distinguished from LawDroid, where the signal is live because Copilot performs case law research. Checked the home page, the AI Receptionist page and the directory listing on 29 Aug 2026.
Refusal and Uncertainty Behaviour
What does the product do when the answer is not in the corpus?
No located public material addresses what the product does when it cannot ground an answer.
Not addressed by the vendor, and third party reporting describes the behaviour that fills the gap. Nothing published states what the AI does when it cannot understand a caller, when a question falls outside the firm's script, or when an intake condition is ambiguous. What independent review reports is that the AI Receptionist auto escalates calls to live agents without user authorisation, which is in substance an uncertainty behaviour: the system hands off when it cannot proceed. That is a reasonable design and arguably the right one for a distressed caller, and it is undisclosed, ungoverned by any published rule, and carries a per call cost the customer did not authorise. A behaviour a customer only discovers through their invoice is not a disclosed behaviour. Checked the home page, the AI Receptionist page, the pricing material and independent review on 29 Aug 2026.
Fabricated Citation Record
Does a public court record exist involving output from this product?
No court order, opinion or disciplinary record naming this product has been located as of the date shown. This is a statement about the public record, not a finding about the product.
None located, with the instrument named. General web searches combining the vendor name with court, order, sanction, complaint and consumer protection terms returned nothing on 29 Aug 2026, and no named docket database, court record tracker or state consumer protection register was searched. Recorded as a statement about what this search found, not as a clearance. The exposure shape is not fabricated citations, since no legal authority is generated. The analogous adverse findings would be call recording consent under state wiretapping statutes, or a billing dispute arising from the auto escalation pattern independent review reports, and neither would surface through the search terms a citation focused instrument uses. Flagged for a differently scoped search on a later pass.
Bar Guidance Alignment
Has the vendor engaged in public with the ethics opinions its buyers are bound by?
No located public material engages with bar or ethics guidance.
Not addressed. No named ethics opinion, no ABA Formal Opinion 512, no state bar guidance and no engagement with professional conduct rules was located. The relevance is direct and long standing rather than novel: outsourced legal intake by non lawyers has been the subject of bar guidance for years, covering supervision of non lawyer assistants, confidentiality of prospective client communications and fee sharing, and this vendor is the dominant provider in that market with agents trained on attorney style intake scripts. A vendor that has built its brand on legal intake and engages none of the rules governing it is a more conspicuous omission than the same silence from a general purpose tool. Third of three records in this category at this value. Checked the home page, the AI Receptionist page, the directory listing and the site navigation on 29 Aug 2026.
Billing and Fee Posture
Does the vendor address what happens to the bill when the work takes an hour instead of six?
Public materials claim time savings without addressing billing or disclosure.
Savings claims only, and the cost comparison is unusually concrete. The vendor's published argument is a staffing substitution: it states that an in house nine to five weekday receptionist costs $4,000 or more per month with months to ramp, constant turnover and limited coverage, against published plans starting far below that with 24/7 coverage. Naming the alternative's cost and its non financial drawbacks makes the claim checkable in a way a percentage never is. Customer outcome claims are attributed to named people at named firms and carry no methodology, including a 60 percent year over year sales profit increase. Nothing appears on the client's side of the equation: no position on whether outsourced intake cost is passed to clients or absorbed, no guidance on disclosing to a caller that they are speaking to a third party service or to an AI, and no record showing which portion of an intake was machine handled.
Outside Counsel Guideline Readiness
Can a firm get this vendor through a client’s AI clause without a bespoke negotiation?
No located public material supports a client side disclosure obligation.
Not addressed. No trust centre, security page, named certification, subprocessor list, named model or speech provider, data processing agreement or documentation request route was located, so a firm has nothing to forward and no destination to point a client toward. The gap is heavier here than for the software peers in this category because the vendor is not only a processor but a service provider whose human agents handle the firm's client communications, so a client asking who speaks to prospective clients on the firm's behalf, where they are, and how the recordings are held cannot be answered from any public material. Third of three records in this category at this value. Checked the home page, the AI Receptionist page, the pricing material, the directory listing and the site navigation on 29 Aug 2026.
Court Disclosure Support
If a judge’s standing order requires an AI disclosure, can the product produce one?
Some elements of the record are available, short of a document level export.
Partial record, and the first record in this category with a genuine evidentiary artifact. Call recordings and transcripts exist as a matter of product function, conversations are stated to sync into the firm's CRM in real time, and the vendor's own home page demonstrates that recordings are retained and playable, so a firm can in principle produce what a caller said and what was said back, which is more than Gideon or LawDroid can evidence. The relevant forum in this category is a bar complaint, a fee dispute or a malpractice claim rather than a filing, and a recording is exactly the artifact those proceedings turn on. The gaps are the familiar two plus one specific to the hybrid model: nothing indicates that a record marks which portions of a call were handled by AI and which by a live agent, which is the distinction that would matter if what was said to a prospective client were ever contested, and no export or retention artifact is described as available to the customer on demand.