Between the demo close and the pilot kickoff sits a stage many dental AI deals lose silently: the clinical evaluation. The lead hygienist, the lead clinical assistant, or — on multi-location DSO deals — the Director of Clinical Operations arrives in the buying committee and stops asking the questions the practice manager asked. They want to hear how the AI receptionist handles a lost crown call at 9 PM. They want to see what the clinical record looks like the morning after a post-extraction bleed call. They want to know whether the AI knows the difference between a cracked filling that can wait and a cracked filling that cannot. Competitor A (the horizontal voice-AI platform) and Competitor B (the dental-pure AI receptionist) attack this stage on completely different vectors, and the demo script that won the prior meeting is the wrong artifact for the room. This is the clinical-evaluation battlecard that closes the stage cleanly enough to advance to the pilot scorecard with the clinical lead's signal recorded in writing.
TL;DR
Five clinical domains, two attack patterns, one battlecard. A-slot horizontal platforms attack on NLU breadth — "we handle any vertical's clinical-intent calls" — and lose on dental clinical vocabulary depth the moment a real after-hours transcript is played. B-slot dental-pure receptionists attack on dental marketing-copy specificity and lose when their actual clinical-intent transcript reveals keyword-match routing with no triage logic in between. The Voicify rep plays a live after-hours emergency transcript, walks the clinical lead through the dental clinical-intent taxonomy, shows the clinical-record write-back diagram, and never runs the clinical-evaluation meeting solo — the Voicify clinical lead is on every call where a competitor is named as a finalist.
Why the Clinical Evaluation Is Its Own Stage
The clinical evaluation is its own stage because the buying committee changes and the decision criteria change with it. The practice manager who drove the economic case in the discovery brief and the demo is now in observer mode. The clinical lead is in primary, often with a senior hygienist or the practice owner-dentist copied in. Their criteria are not the criteria that won the demo. They do not care about ARR or per-call cost; they care about whether the AI knows that a patient who calls and says "my temporary crown fell off last night and I have a porcelain crown seated next Tuesday" should be routed to next-day scheduling with a note to the assigned dentist — not voicemail, not same-day emergency, and not generic scheduling. The master battlecard treats the clinical evaluation as a single stage but does not break out the five clinical domains in depth because the master battlecard is written for the rep who needs the top-level slot and stage map. The clinical-evaluation battlecard is the layer below.
It also fires on its own trigger. The trigger map treats "clinical lead joins the buying committee as primary" as a named signal in its own right — separate from the IT-joins signal, separate from the pricing-escalation signal — because the artifact, the activation cue, and the role transition are all different. The AE does not own the clinical-evaluation stage. The Voicify clinical lead does, with the AE in support and the product team on standby for the clinical-intent taxonomy walkthrough. The activation cue is unambiguous: the moment the clinical lead joins the committee in primary, the Voicify clinical lead is on the next call. The AE who runs the clinical evaluation solo creates a silent-veto risk no forecast call can recover.
The Five-Domain Clinical Evaluation Surface
The clinical evaluation covers five domains. Each domain has a slot-specific attack pattern from Competitor A or Competitor B, a Voicify response, and the live-call transcript or clinical-intent reference the rep or clinical lead plays in the meeting. The five domains are exhaustive in the sense that questions outside them are rare; they are not exhaustive in the sense that any one domain can sink the deal alone.
| Domain | A-slot attack | B-slot attack | Voicify anchor |
|---|---|---|---|
| Clinical-intent taxonomy | "Any vertical, configurable NLU" | Keyword-match routing on dental terms | Dental clinical-intent taxonomy with named escalation paths |
| Emergency triage | Generic urgency scoring, no clinical hierarchy | "Emergency" keyword routes to voicemail | Dental-specific triage logic with on-call-dentist escalation |
| Post-op call handling | No post-op-specific flow | Generic callback message, no clinical context | Procedure-aware post-op flow with chart-noted intent |
| Clinical-record write-back | Read-only sync, no clinical-note write | Scheduling-only fields, no clinical context | Chart-noted intent in Dentrix/Open Dental clinical notes |
| Clinical voice & tone | Generic empathic voice, no pain-context shift | Friendly but flat, no triage register | Pain-context-aware tone shift with dental empathy script |
Five rows, two attacks per row, one Voicify anchor per row. The clinical lead does not read the table the way the AE does — they read it as a list of behaviors they will probe with specific clinical scenarios. The rep's job is to anticipate the probes, pre-stage the transcripts that demonstrate the Voicify anchor in each domain, and put the Voicify clinical lead in the room to answer the dental-specific follow-ups when they come.
Clinical-Intent Taxonomy: The Anchor Domain
Clinical-intent taxonomy is the anchor domain because it is where the A-slot horizontal platform's general-purpose NLU cannot follow Voicify into dental clinical specificity. The horizontal platform supports a configurable intent taxonomy in the marketing copy — define your verticals, train on your transcripts, deploy your prompts. In practice, the dental clinical-intent taxonomy is not a list of twenty intents the customer can configure; it is a hierarchy of more than two hundred dental clinical scenarios with named escalation paths — lost crown, lost temporary crown, lost permanent crown post-cementation, cracked filling without pain, cracked filling with cold sensitivity, cracked filling with spontaneous pain, post-extraction bleed first 24 hours, post-extraction bleed 24 to 72 hours, post-extraction bleed beyond 72 hours, post-op pain on a recently completed root canal, post-op pain on a recently seated implant, dry socket symptomatology, abscess with facial swelling, abscess without facial swelling, and so on. Each scenario has a named routing path, a named escalation criterion, and a named on-call protocol. The horizontal platform's general-purpose NLU does not have this depth because the architecture was not built for it; the customer cannot configure it in because the underlying clinical vocabulary training data does not exist.
The Voicify response is the clinical-intent taxonomy walkthrough. The Voicify clinical lead hands the lead hygienist the dental clinical-intent reference — a printed or shared-screen document that names every clinical scenario the receptionist handles, the routing path for each, and the on-call-dentist escalation criterion. The B-slot dental-pure competitor sometimes has a partial version of this — usually fifteen to thirty named clinical scenarios — but the rest collapse into a single "emergency" bucket that resolves to voicemail with a callback message. Walking the B-slot scenario list line by line with the clinical lead surfaces the gap the marketing copy hid.
Emergency Triage and Post-Op: The Two Domains Reps Underweight
Emergency triage is the domain the clinical lead probes hardest because it is the domain a clinical lead's reputation rides on. A receptionist that routes a post-extraction bleed to voicemail with a "we will call you back in the morning" message creates a liability the clinical lead's name is attached to. The A-slot horizontal platform applies generic urgency scoring — high, medium, low — with no dental clinical hierarchy underneath. The B-slot dental-pure competitor matches on the keyword "emergency" and routes to voicemail. Voicify applies dental-specific triage logic: a post-extraction bleed in the first 24 hours after a third-molar extraction escalates to the on-call oral surgeon directly; a post-extraction bleed beyond 72 hours routes to next-day scheduling with a chart note for the assigned dentist to review at the appointment; a cracked filling with cold sensitivity routes to next-available scheduling; a cracked filling with spontaneous unprovoked pain escalates to same-day scheduling with an emergency carve-out. The clinical lead probes this logic by reading the receptionist three or four real after-hours scenarios from the practice's last quarter and listening to how the AI routes each. Voicify's clinical lead is in the room to walk the routing live; the AE running the meeting solo cannot.
Post-op call handling is the domain reps most often skip and clinical leads most often gate on. A patient who had a crown seated last Tuesday and calls Friday night with sensitivity is not a generic clinical-intent call; the receptionist needs to know the procedure, the date, the assigned dentist, and the typical post-op timeline before routing the call. The A-slot horizontal platform has no post-op-specific flow because the patient-procedure-history join does not exist in the platform's data model. The B-slot competitor sometimes has a generic post-op callback message but no procedure-aware logic underneath. Voicify writes a procedure-aware post-op flow that pulls the patient's recent procedure history from Dentrix or Open Dental, applies the procedure-specific post-op timeline, routes the call accordingly, and writes a chart-noted intent record back to the patient's clinical notes. The clinical-record write-back is what the clinical lead will look at the morning after the call to evaluate whether the AI captured the right context. The IT due-diligence battlecard covers the PMS write-back at the integration layer; this battlecard covers what gets written from the clinical-intent perspective.
Clinical Voice and Tone: The Domain No One Documents
Clinical voice and tone is the domain that decides the evaluation for clinical leads who run their own front desk and have heard every variation of how a patient in pain sounds at 11 PM. A receptionist whose voice does not shift register when the caller is in pain — whose tone stays in scheduling-friendly mid-register the entire call — is a receptionist the clinical lead will not put on the after-hours line. The A-slot horizontal platform's voice model is generically empathic but does not shift context-aware register based on the dental clinical-intent detected. The B-slot dental-pure competitor's voice model is friendly but flat — appropriate for an appointment-confirmation call, inappropriate for a patient with a swollen face. Voicify trains the voice model on dental pain-context registers with explicit tone shifts triggered by clinical-intent detection: scheduling-friendly mid-register for routine calls, lower-pace empathic register for moderate pain, slow-pace acknowledging register for severe pain or emergency. The clinical lead probes this by listening to two or three pain-context calls back to back with the rep in the room. Voicify's clinical lead identifies the tone-shift markers in real time, which a horizontal-platform AE cannot do because the markers are not documented in the platform's voice-config spec.
The Five Clinical Stop-Ship Probes
Five probes stop deals at the clinical-evaluation stage more than any other. The rep and the Voicify clinical lead should rehearse these five answers as a unit before the clinical-evaluation meeting, with the clinical-intent taxonomy open and the after-hours emergency transcripts queued for playback. Each probe is a clinical-lead question anchored on a specific transcript or taxonomy reference the Voicify clinical lead can play in the same meeting.
- "Play me a real after-hours emergency." — Cue the post-extraction bleed transcript with the on-call-dentist escalation. A-slot does not have a dental-specific transcript; B-slot has a transcript but the escalation resolves to voicemail.
- "What does the chart note look like the morning after?" — Show the Dentrix or Open Dental clinical-note write-back from the prior night's call. A-slot does not write clinical notes; B-slot writes scheduling fields only.
- "How does it know a lost temporary crown from a lost permanent crown?" — Walk the dental clinical-intent taxonomy at the named-scenario level. A-slot resolves both to a generic "lost crown" intent; B-slot does the same.
- "What happens on a post-op call after a same-day implant?" — Walk the procedure-aware post-op flow with the patient-history join from the PMS. A-slot has no post-op flow; B-slot has a generic callback message.
- "Play me the voice on a patient in severe pain." — Cue the severe-pain register transcript and identify the tone-shift markers. A-slot voice does not shift register; B-slot voice is flat across intents.
The pattern is consistent across all five: A-slot has the capability framing at platform scope but not at dental clinical depth; B-slot has dental marketing copy but lacks the underlying clinical-intent depth. Voicify wins the clinical evaluation when the Voicify clinical lead plays every transcript and walks every reference in the same meeting the clinical lead probes.
How the Clinical Battlecard Hands Off to the Pilot Scorecard
The clinical evaluation closes when the clinical lead signals — verbally in the meeting, then in writing in a follow-up email or CRM note — that the receptionist behavior meets the practice's clinical-intent standard for pilot. The activation cue for the hand-off into the pilot scorecard is that written signal — usually a short email from the clinical lead to the practice manager with the AE copied that names the five clinical domains evaluated and the clinical lead's approval to proceed. The AE does not advance the deal into pilot kickoff before that email exists. Reps who attempt the pilot kickoff while the clinical lead is still in clinical evaluation create the silent-veto risk that surfaces three weeks into the pilot when the clinical lead's pilot scorecard is filled out with sub-threshold marks the AE cannot now appeal. The handoff into the objection-handling playbook happens any time the clinical lead's probe surfaces a real gap the rep cannot close in the meeting; the slot decision may need to be revisited if the clinical-evaluation transcripts reveal the competitor is operating in a slot the rep had not previously assigned. The clinical evaluation is short — one to two meetings — but the signal it produces is load-bearing for every stage that follows, and the battlecard makes sure the signal is captured in writing and not in the AE's notebook.