The slots themselves live in your PMS — the AI reads them. What you configure in the AI dashboard is the interpretation layer: how the AI maps patient requests to your slot types, which providers see what, how same-day urgency is handled, and what your practice's unwritten rules are ("Dr. Patel doesn't see new patients before 10am"). Once configured, the AI respects all of it without manual intervention.
Here's what that setup looks like in practice and the tuning decisions that make the biggest difference.
What the PMS Already Tells the AI
Your PMS is the source of truth for:
- Provider schedules by day and time
- Operatory / chair assignments
- Appointment types with configured durations (30-min cleaning, 90-min crown prep, etc.)
- Block scheduling (new-patient blocks, hygiene-only blocks)
- Holidays and closures
- Provider vacation
You don't duplicate any of this in the AI. The AI queries live and books against whatever is currently true.
What You Configure in the AI Dashboard
1. Appointment-type taxonomy mapping
Patients say things like "I need a cleaning" or "my tooth hurts." Your PMS has appointment types like "RECALL_ADULT" and "EMERGENCY_LIMITED_EXAM." The mapping translates the patient's words to your PMS codes. Mapping is done once at setup, takes 30 minutes, and is refined occasionally.
2. Routing rules per appointment type
For each type, who sees it? New-patient exams might go round-robin among 3 providers; hygiene goes to hygienists; implant consults go only to Dr. Kim. Codified from your existing unwritten rules.
3. Same-day capacity and emergency slots
Most practices hold a few slots per day for emergencies. Configure how the AI uses them — reserved for true emergencies only, available for any urgent request, or shared with walk-ins. Policy choice, not AI limitation.
4. Minimum scheduling window
How far in advance can a patient book? "No earlier than 1 hour from now" is typical. Prevents booking someone for the same slot they're already in.
5. Maximum scheduling window
How far out can a patient book? "6 weeks" is common; some practices set longer for hygiene recalls.
6. Patient-specific rules
Chronic no-show patients: require deposits or restrict to walk-in. High-value patients: preferred-provider routing. New patients: intake form requirements before the slot confirms.
7. Provider preferences
"Dr. Patel sees crowns Tuesdays and Thursdays only." "Dr. Kim doesn't see new patients before 10am." "Dr. Lee doesn't take implant consults." These unwritten rules live in the PMS where possible; the AI layer catches what the PMS can't represent.
Common Tuning Decisions After Launch
Should the AI offer 2 or 3 slot options?
Two usually converts better (less decision fatigue). Three feels more generous but can cause hesitation. Default to two; test three if conversion lags.
Should same-day slots appear in standard booking, or only emergencies?
Practice preference. Reserving them for emergencies keeps them for the patients who need them. Making them generally available fills more seats but sometimes leaves emergencies without room.
Morning vs. afternoon defaults
Some practices want the AI to skew toward filling morning slots (historically harder to fill). Configurable.
New-patient routing
Round-robin, single-dedicated-provider, or skew toward newer associates to build their books. Review monthly for the first quarter.
Handling last-minute requests
If a patient asks "can I come in today?" and no slot fits, does the AI offer a next-day morning slot proactively, or simply say "nothing available today"? Configure for your preference.
What Not to Configure
- Don't duplicate PMS rules in the AI. Single source of truth is the PMS.
- Don't try to capture every edge case upfront. Launch with 80% coverage and refine. The first 30 days' call data will show you what's missing.
- Don't over-engineer same-day policy. Simple rules ("hold 2 slots per provider per day for emergencies") work better than complex ones.
Monthly Review Cadence
- Review appointment-type mapping accuracy: are patients being routed correctly?
- Check provider distribution: are new patients distributed as intended?
- Look at same-day fill rate: are emergency slots used or wasted?
- Identify missed mappings: patient requests that didn't match any configured type
Fifteen minutes a month is usually enough after the first quarter.
FAQ
What if my PMS doesn't have granular appointment types?
Most modern PMSs do. If yours doesn't, the AI layer can add the taxonomy and book into generic slots. Less clean but functional.
Can the AI suggest slot-policy changes based on patterns it sees?
Some vendors have analytics that show "35% of your new-patient requests are outside your current new-patient windows." The decision to expand the window is yours; the AI surfaces the pattern.
How do I handle providers who don't follow my configured rules?
Clean rules in the PMS and AI prevent this. If Dr. Patel claims he'll "see crowns on Mondays if needed," make that explicit in the rule ("occasional exceptions allowed, default Tuesday/Thursday"). The AI can handle tiered preferences.
Can I block booking for specific slots ad hoc?
Yes — your PMS's standard "appointment block" feature. The AI reads it in real time.
What happens to booked slots when the schedule changes?
If a provider's day gets cancelled, the AI can proactively call affected patients to reschedule. Your PMS triggers the outbound list; the AI does the calls.