Intiveo

A mature messaging and engagement platform that added two model capabilities two months before this assessment, and the company is candid that this is the beginning rather than the substance.

Everything that made this business is deterministic. Two way texting, recall reminders, appointment confirmations, waitlist automation, referral management, surveys, reputation management and integrated voice are messaging infrastructure and workflow. The waitlist feature, which the company's own benchmark data credits with recovering an average of 2,773 appointments a month across participating practices, is rule driven matching of cancellations to waiting patients. That is the most commercially valuable automation on the platform and no model is involved.

Veo, launched 25 June 2026, ships two capabilities. Suggestions recommends a chat template matching a patient conversation, which requires classifying what the patient is asking about and ranking templates against it. Responses answers common inquiries outside business hours on topics including new patient intake, office location and parking. Both are genuine model work and both are narrow: one recommends a template a human sends, the other answers logistics questions when nobody is there.

The framing is honest and worth crediting rather than discounting. The company describes Veo as a growing network and the launch as the first step in a long term investment, positions the capabilities around control and configurability, and its chief executive states plainly that practices are not looking for technology that removes the human element.

Graded D, alongside Yosi Health. A product whose artificial intelligence suggests message templates and answers parking questions is not a model business, and the company does not claim to be one.

The most carefully bounded automation design in this session, and the boundaries are stated in the launch material rather than discovered afterwards.

The two capabilities are separated by exactly the right variable, which is whether a human is available. During business hours Veo Suggestions recommends a chat template and the staff member sends it, customises it or dismisses it, so three outcomes are described and two of them involve the human overriding the model. Outside business hours Veo Responses answers autonomously, and the scope is restricted to inquiries with no clinical content: new patient intake questions, office location, parking. Automating parking directions at eleven at night is close to the safest possible application of a language model in healthcare.

The framing around it is consistent and unusually explicit. The company describes Veo as configurable so practices adopt at their preferred pace, positions the programme around control, transparency and trust, and its chief executive states that practices are not looking for technology that removes the human element from care. Vendors routinely say the opposite while shipping something more autonomous than they admit; here the modesty of the claim matches the modesty of the capability.

Two gaps hold this at B rather than A. Nothing describes what Responses does when an after hours message is not routine, and a dental practice receives out of hours messages about pain, swelling and post extraction bleeding, which are the messages that matter most. No escalation path, urgency detection or emergency instruction behaviour is described. And nothing states whether patients are told they are receiving an automated reply.

Graded B, the highest on this axis in this session.

Behaviour is described precisely, mechanism is not described at all, and one upstream dependency is named which is more than most records manage.

On behaviour the launch material is clear. Suggestions recommends chat templates matching a patient conversation during business hours, with three stated user actions: send, customise, dismiss. Responses is configured by the practice to answer named categories of inquiry outside business hours, with the categories enumerated as new patient intake, office location and parking. A buyer knows exactly what each feature does and where the boundary sits, which is better than the agent labelling on the neighbouring Clearwave record where eight capabilities share one word.

The named dependency is telephony. Voice is described as built on a commercial cloud communications provider, identified by name, characterised as a global leader in artificial intelligence powered business communications. Naming the infrastructure partner is a real disclosure and it tells a buyer whose network carries their patient calls.

Below that, nothing. No architecture, no language model provider for Veo, no training data description, no accuracy figures for template matching or automatic responses, no evaluation method, no versioning and no update cadence. Veo is described as a growing network of capabilities, and nothing states how a practice learns when a new capability arrives or when an existing one changes behaviour.

That matters more for the after hours responder than for the suggester, since a change in what it will and will not answer alters what patients receive unsupervised.

Graded C.

One infrastructure dependency is named clearly, the model dependency is not, and the naming that does occur is more than most records in this session offer.

The named dependency is telephony. Voice is described as bringing together cloud based phone communication from a specifically identified commercial provider with the platform's engagement features, and that provider is characterised in the company's own material as a global leader in artificial intelligence powered business communications. Naming the partner tells a buyer whose infrastructure carries their patient calls and where a subprocessor sits, and it is disclosed in a product announcement rather than extracted from a contract.

The practice management integration side is partly visible too, with an open source dental system named specifically.

Veo is where the disclosure stops. Two model capabilities launched in June 2026 by a company of this size, and nothing states whether the underlying language capability is built in house, licensed, or running on a commercial foundation model. The presumption for a company at this scale runs strongly toward integration, and template matching plus generated responses to patient inquiries is exactly the workload a commercial model service handles. Each unnamed provider is a party receiving patient chat content.

That gap sits awkwardly against the launch positioning. Veo was announced as a trusted, transparent approach to artificial intelligence, and the single most transparent thing a vendor can say about a model feature is what the model is. The company named its telephony partner and did not name this one.

Graded C on the strength of the disclosure that does exist.

The evidence here is unusual in this index because the company publishes operating data rather than marketing claims, and it publishes numbers that are not flattering.

The July 2026 benchmark report draws on communication and engagement data from more than 2,500 practices over a twelve month period from March 2025 to February 2026, segmented by practice size. The figures are specific and checkable in kind: confirmation rates at 77 percent industrywide with text driving about two thirds of confirmations, voice growing from 2 percent of channel mix to more than 7 percent, referral conversion to completed treatment at 16.1 percent for solo practices, 23.7 percent for multi location and 25.5 percent for enterprise, and waitlist users recovering an average of 2,773 appointments a month. A vendor publishing that solo practices convert referrals at two thirds the rate of enterprise groups is publishing something its smallest customers will not enjoy reading, which is a sign the numbers came from the data rather than from marketing.

Adoption evidence is independent in one respect that matters: the platform is described as used in 85 percent of dental schools in North America, and academic adoption is decided by faculty rather than by a purchasing committee responding to a sales process.

Customer material is qualitative, with named practitioners and specific operational claims such as three hours saved a day.

Two qualifications. The benchmark data comes from the company's own customer base, so it describes practices using this software rather than the dental market, and the report is a marketing asset as well as a research one. And there is no evidence whatever for Veo, which is two months old.

Graded B on the strength and specificity of the operational data.

No stewardship statement exists, and one disclosed practice sits uncomfortably close to the question without answering it.

That practice is the benchmark report. The company analysed communication and engagement data across more than 2,500 practices over twelve months and published aggregate findings. That is secondary use of customer data, it is disclosed openly rather than hidden, and the output is genuinely useful to the industry. It also demonstrates that customer data is pooled and analysed centrally, which establishes the capability and the willingness without establishing the terms. Nothing published states whether practices consented, whether they can opt out, what de identification was applied before aggregation, or whether the same pooled data informs Veo's models.

That last question is the live one. Veo suggests chat templates by matching them to patient conversations, and the obvious training material for such a model is the conversation history of 3,000 practices. Whether Veo learns from customer messages, and whether learning is confined within a practice or pooled across the base, is unstated in either direction, and the company launched Veo on a platform of transparency and control without addressing it.

The content at stake is moderate rather than severe. Chat conversations with a dental practice contain appointment logistics, occasional symptom description and payment discussion, and voice communications add audio. Sensitive, and thinner than the intake records on adjacent pages.

One design point works in the vendor's favour: Veo Suggestions puts a human between the model and the patient by default, so the model reads conversations rather than conducting them during business hours.

Graded C.

One certification is published and the cross border question that is distinctive to this record is unaddressed.

The company announced successful completion of a service organisation control type 2 audit, which tests controls over a period rather than at a point, and it was announced as a company milestone rather than buried in a footer. For a vendor of this size that is a real investment.

The data flows are narrower than on the neighbouring intake records and that works in the vendor's favour. This platform carries appointment details, reminders, chat conversations, survey responses, referral information and voice communications. It is not collecting insurance cards, government identification or clinical intake histories, and no evidence was located that it processes card payments. The protected information exposure is real and it is thinner.

The distinctive gap is jurisdictional. The company is headquartered in Vancouver, serves practices across North America, and the majority of its market is in the United States. That means patient information from United States dental practices is handled by a Canadian company, engaging cross border transfer questions and Canadian privacy law alongside the United States health privacy statute. Nothing located addresses where data is stored, whether United States customer data stays in the United States, or how the two regimes are reconciled. No other record in this session raises this question and it is the first thing a United States practice's counsel should ask.

No business associate agreement is offered or described, no protected data handling summary exists and no retention position is stated, including for chat transcripts and voice communications.

Graded C.

One real certification, announced as a milestone, with nothing built around it.

The company states successful completion of a service organisation control type 2 audit and treated it as news rather than as a footnote, which for a company of this size represents genuine investment. Type 2 tests whether controls operated over a period rather than existed on a date, so it is the more meaningful of the two report types.

That is where it ends. No trust centre, no described process for requesting the report under agreement, no audit period or auditor named, no scope statement, no penetration testing reference, no vulnerability disclosure policy, no subprocessor list and no incident notification commitment were located. A prospective customer has a claim and no route to verify it.

Scope is worth asking about specifically here because of timing. The audit predates Veo, which launched in June 2026, and it predates the voice product built on an external communications provider. Whether the certified boundary now includes the artificial intelligence capabilities and the telephony integration is unstated, and on a platform adding capabilities as a growing network that boundary will keep moving.

The comparison within this session is instructive. Yosi Health, a similarly sized company, publishes four credentials covering both the clinical and payment halves of its product. Intiveo publishes one, which is proportionate to a narrower data estate that handles no payment card data and no clinical intake, and it is still a single claim with no apparatus behind it.

Graded C.

No clearance is claimed and none is required. Appointment reminders, chat, surveys, referrals and reputation management carry no clinical claim, and Veo suggests message templates and answers questions about parking. Nothing here approaches a device question and the company does not overclaim.

The regimes that apply are communications and privacy, and both are more complicated for this vendor than for its United States peers because it operates across a border.

On communications, a platform whose core function is outbound texting, voice reminders and recall campaigns operates squarely inside telephone consumer protection rules in the United States, covering consent, revocation and calling windows, and inside anti spam legislation in Canada, which is stricter than its United States equivalent on consent for commercial electronic messages and carries meaningful penalties. Recall reminders sit close to the line between transactional and promotional, and that line is where enforcement happens. Nothing published addresses consent capture or opt out handling in either jurisdiction.

On artificial intelligence disclosure, the after hours Responses capability replies to patients automatically, and several jurisdictions now require that a person be told when they are interacting with an artificial agent. Nothing states whether the reply identifies itself as automated.

Dental practice regulation itself is state and provincial, and reputation management touches advertising rules that dental boards enforce, another area nobody discusses.

Graded C: correctly outside device regulation, silent on the communications regimes that are its actual operating environment.

Nothing was located. No model card, no training data description, no accuracy figures for either Veo capability, no subgroup analysis and no bias statement.

The absence is more conspicuous here than the small scope of the capabilities suggests, because the company launched Veo explicitly on a platform of transparency. The announcement uses the words transparent, configurable and trust, and describes the programme as a trusted approach to artificial intelligence built for dental practices. A vendor making transparency the product positioning invites the question of what it has actually disclosed, and the answer is a description of what the features do and nothing about how they work or how well.

The substantive exposure is language. Dental practices serve linguistically diverse populations, template suggestion depends on interpreting what a patient wrote, and after hours automatic responses reply to whoever messages. A patient writing in imperfect English, or in another language entirely, may receive a mismatched template or an unhelpful automatic reply at the moment when no human is available to correct it. Nothing published describes language coverage for either capability.

A second and smaller mechanism sits in template suggestion itself. A model that learns which templates staff send will reproduce existing communication patterns, including any differences in how practices have historically communicated with different patients.

The scope genuinely limits the harm. Sending a slightly wrong template about an appointment is recoverable. Graded D on the absence rather than on the severity.

Nothing published addresses responsibility for an automated outcome, and this record grades above the session floor because the design limits the exposure rather than because the vendor has addressed it.

The structural mitigations are real. During business hours a human sends every message, so the model cannot communicate with a patient unsupervised. Outside hours the automatic responder is scoped by the practice to named categories that contain no clinical content. A vendor that has confined its autonomous surface to parking and office location has, whether deliberately or not, confined its liability surface too.

The unaddressed case is the one that scope does not cover. A dental practice receives out of hours messages about pain, swelling, bleeding after an extraction and lost temporary crowns. If a patient describes one of those and the automatic responder replies with something about office hours or new patient intake, the patient may reasonably read that as the practice having responded and wait until morning. Nothing published describes urgency detection, an escalation path, an emergency instruction fallback, or what the responder does with a message it was not configured for. That is the gap, and it is a foreseeable rather than exotic scenario.

Secondary exposures are commercial. Reputation management touches dental board advertising rules. Recall and reminder messaging carries telephone and anti spam consent exposure in two countries, and the platform sends on the practice's behalf so the practice is the sender of record.

No indemnity, limitation, performance warranty or recourse route was located, and no error rate is published for either Veo capability.

Graded C.

Adequate for the segment, thin against the rest of this session, with one integration choice worth crediting.

Dental practices run practice management systems rather than hospital record systems, and the platform integrates with them, with an open source dental practice management system named specifically and its cloud offering supported. Naming a specific system with a specific deployment model is more concrete than a generic integration claim, and choosing to support an open source platform reaches independent practices that proprietary only vendors miss.

The referral capability is the strongest interoperability function on the record and it is organisational rather than technical. Two way referral management moves a patient between a general practice and a specialist, which in dentistry is a genuine coordination gap usually handled by fax and phone. A communication feature extending to caregivers, extended family and referring practices addresses the same gap from the other direction. The benchmark data quantifies why it matters, reporting referral conversion to completed treatment at 16.1 percent for solo practices against 25.5 percent for enterprise groups.

What is absent is breadth and specification. No count of supported practice management systems was located, no application programming interface is published, no interoperability standard is named, and nothing describes what data moves in which direction or whether clinical information transfers at all. Against Yosi Health on the adjacent record, which names six record systems each with its own integration page and writes discrete data into the chart, this is materially less documented.

Graded C.

Nothing published, and this is the one record in this session where residency is a live question rather than a formality.

The company is headquartered in Vancouver, British Columbia, and serves more than 3,000 practices across North America with the majority of that market in the United States. So patient communication data from United States dental practices is being handled by a Canadian company. That engages cross border transfer considerations, Canadian federal and provincial privacy law alongside the United States health privacy statute, and the practical question of which jurisdiction's authorities could compel access to the data. British Columbia in particular has historically had strict public sector data residency rules, and while those apply to public bodies rather than to this vendor's customers, the jurisdictional picture is genuinely more complicated here than for a domestic vendor.

Nothing published addresses any of it. No hosting platform, no region, no statement of whether United States customer data is stored in the United States, no residency option, no subprocessor list, no retention position and no export or contract end terms.

One dependency is partly visible and points outward again. Voice runs on a named commercial cloud communications provider, so patient calls traverse that provider's infrastructure, and nothing describes where.

Availability is unaddressed, with no uptime commitment or status history, which matters less here than for an intake platform gating a waiting room but still governs whether after hours automatic responses actually respond.

Graded D on the absence, with the cross border point recorded because no buyer should have to discover it themselves.

No figures are published and the packaging is clearer than most, which lifts this one band above the floor.

What is visible is structure. The company sells distinct products, naming an entry level offering described as making patient engagement more accessible to practices of all sizes, alongside multi location and enterprise or dental service organisation tiers and an academic programme. Publishing a named entry tier tells a solo practitioner that there is a product for them, which is the single most useful thing a small buyer needs to know before starting a conversation, and it is more than any of the intake vendors on the adjacent records offer.

Voice is built on a commercial cloud communications provider, which implies telephony costs sit somewhere in the arrangement, and nothing describes whether they are bundled or passed through.

No rate card, unit of charge, contract term, minimum or implementation fee was located. For a platform sold to solo practices through to thirty location groups, the unit is almost certainly per location or per provider with tier gating, and it is not stated.

One question is specific to the timing. Veo launched in June 2026 as a growing network of capabilities, and nothing states whether it is included in existing subscriptions or will become a paid add on as it expands. Customers signing now have no published basis for knowing whether the artificial intelligence they are being shown is part of what they are buying.

Graded C.

Deliberately narrow by specialty and genuinely complete within it, which is the correct trade for this product and produces a middling grade against an index of broader vendors.

The specialty is dental and oral surgery, and the depth within it is real rather than nominal. Features are described as specialty specific, oral and maxillofacial surgery has its own material, and the benchmark report segments dentistry rather than treating it as one market. A referral management product connecting general practices to specialists reflects how dental care actually flows, and a communication feature extending to caregivers, extended family and referring practices reflects who is actually involved in a dental episode.

Organisational coverage inside dentistry is complete: solo practitioners, multi location groups, dental service organisations up to thirty locations, and an academic segment reaching 85 percent of North American dental schools. Serving a solo practice and a thirty location group from the same platform is real range, and the academic footprint means a substantial share of new dentists train on this software, which is a durable distribution advantage.

Geography is North America, with the company based in Canada, so the platform operates across two regulatory regimes.

What holds this at C is that the whole record covers one specialty and one that sits at the edge of this index. Against vendors serving every service line in a health system, a dental only platform is narrow by definition, and no medical, behavioural or allied health coverage exists or is claimed.

Graded C.