Un réseau dentaire québécois déflecte le tiers de son volume de centre d'appels avec une réceptionniste IA.
Devpro built a virtual frontline receptionist for 123 Dentiste, a Quebec dental clinic network, handling 1,600 inbound calls per week. The agent identifies whether callers are new or existing patients, determines which clinic they need, and routes accordingly, deflecting 32.5% of all call volume away from the central call centre and directly to the appropriate clinic.
- Engagement
- ~8 weeks · full deployment
- Region
- Quebec, Canada
- Scope
- Front-desk triage · clinic routing
- Volume
- 1,600 calls / week
- Status
- Live in production
Every call hit the call centre, whether it needed to or not.
With 1,600 calls coming in each week across multiple clinics, 123 Dentiste's central call centre was the single point of entry for every patient interaction. The model didn't scale, and patients calling their own clinic to book or follow up were queuing behind calls the call centre couldn't resolve without the clinic anyway.
- No triage at the front line. Every call, regardless of complexity or destination, entered the same queue. A simple appointment inquiry at one clinic competed for the same agent as a multi-location administrative matter.
- Clinics left out of the loop. Calls that could have been handled directly at the clinic level were absorbed by the call centre, adding handling time without adding value and often requiring a callback from the clinic regardless.
- New and existing patients treated identically. New patients and returning patients have different needs and often need to reach different contacts. Without front-line triage, every caller was routed the same way.
Triage at the front. Route to the right place on the first ring.
We designed the agent around the actual decision tree the call centre was applying manually, mapping new versus existing patient paths and per-clinic routing logic before building a single flow.
Discovery
Audited inbound call data to understand volume by clinic, call type, and resolution path. Mapped the routing logic the call centre was applying and identified the cases that could be deflected to clinic direct.
- Call type audit by clinic
- Routing logic mapping
- Deflection opportunity analysis
Build
Built the frontline agent with new versus existing patient identification, per-clinic routing, and integration into the clinic phone routing system. Configured deflection logic and call centre handoff paths.
- Patient type identification
- Per-clinic routing flows
- System integration
- Call centre handoff
Testing
Validated all routing paths across patient types and clinic destinations. Tested edge cases, measured deflection accuracy, and confirmed call centre handoff quality.
- Routing accuracy validation
- Deflection rate testing
- Handoff quality review
Deployment
Deployed as the primary inbound entry point across the network. Post-launch monitoring through the first weeks of full volume.
- Network-wide cutover
- Live monitoring
- Post-launch tuning
A frontline agent that routes correctly before anyone picks up.
The virtual agent answers every call, qualifies the caller, and routes to the clinic or the call centre based on what the patient actually needs, not which queue opened first.
New vs. existing patient identification
The agent establishes on the first exchange whether the caller is a new or existing patient, setting the routing path before any further qualification.
Per-clinic routing
Calls destined for a specific clinic are identified and transferred directly, bypassing the call centre for interactions the clinic can resolve itself.
Call centre handoff for complex calls
Calls that require central handling, multi-clinic coordination, or fall outside the direct routing criteria are transferred to the call centre with caller context already established.
Clinic phone system integration
The agent integrates with the clinic network's phone routing infrastructure, enabling direct transfer to individual clinic lines without manual intervention.
Deflection and volume tracking
Routing outcomes tracked across all call types, giving the network visibility into deflection rate, call centre load, and per-clinic inbound volume on an ongoing basis.
1,600 calls a week. A third no longer touch the call centre.
Measured at steady state post-deployment, against the prior baseline where all inbound volume entered through the central call centre.
Total inbound calls handled by AI frontline
Full network volume, all clinic destinations
Call centre volume reduction
~520 calls per week routed direct to clinic
Weekly call time handled by AI
590 minutes of front-line call handling weekly
Front-line coverage
Calls outside business hours triaged and logged automatically
