50 Dental Practices, One Afternoon — A DSO Platform Uses Mode 3 to Onboard Voice AI Without Touching a Keyboard

The VP of Technology at a regional DSO had done the math. It wasn't complicated, but it was embarrassing.

50 practices. 2 hours per practice to onboard to their voice AI platform. That was 100 hours — two and a half full work weeks — of manual configuration labor for a problem that was, at its core, a data transformation task. Take information about a dental practice, turn it into an agent configuration. Repeat 50 times.

"We had a person whose job was basically to read dental practice websites and fill out forms," he said. "That's not a job. That's something a computer should do."

The Situation

The DSO had been operating a voice AI program for 18 months. The results were clear: practices with AI-answered phones showed a 22% average reduction in missed calls, a 17% improvement in recall acceptance, and meaningful front desk time savings. The program was successful enough that the DSO had committed to rolling it out to all 50 of their practices.

The problem was the onboarding process.

Each practice required:

• A custom system prompt reflecting the practice's specialties, hours, and providers
• A knowledge base built from the practice's insurance network, patient intake process, and service offerings
• Integration setup with the practice management software (Dentrix, Eaglesoft, or Curve, depending on the location)
• Phone number assignment and routing configuration
• A review call with the practice manager to confirm accuracy

Total time per practice: approximately 2 hours, almost entirely manual. The person doing this work was a capable operations coordinator — and she spent the majority of her time doing data entry from dental practice websites.

Scaling to 50 practices would take most of a month. And that assumed no errors, no rework, and no practices with unusual configurations that needed extra attention.

The Approach

Mode 3 is WFW's API-native provisioning mode. Instead of a human operator configuring an agent through a dashboard, the client's platform sends a single API call. Workforce Wave handles the provisioning automatically.

The DSO's engineering team integrated Mode 3 into their practice management platform in approximately three days. The integration was straightforward: when a new practice was added to their platform, or when an existing practice was flagged for voice AI activation, their system sent a single API call to the WFW provisioning endpoint.

The call structure:

POST /v2/agents
{
  "business_url": "https://[practice-website].com",
  "template_id": "dental-general-practice",
  "clientId": "[practice_id_in_dso_platform]",
  "practice_management_system": "dentrix",
  "timezone": "America/Chicago"
}

That was the entire input. The response was an operationId — a reference for an asynchronous provisioning job. WFW's webhook system sent a callback to the DSO platform when provisioning was complete:

{
  "event": "agent.provisioned",
  "operationId": "op_abc123",
  "clientId": "[practice_id]",
  "agentId": "agent_xyz789",
  "phoneNumber": "+18005551234",
  "status": "live",
  "provisionedAt": "2026-02-14T14:23:11Z"
}

The DSO platform received that webhook and automatically updated the practice record with the assigned phone number. No human involvement at any step.

Workforce Wave handled everything between the API call and the webhook: crawled the practice website, extracted specialties, insurance networks, hours, and provider information, built the knowledge base, configured the scheduling prompts, set up the Dentrix integration using the practice management credentials stored in the DSO platform, and assigned a local phone number from the available pool.

The Configuration

The dental-general-practice template provided the base configuration: recall conversation flow, insurance verification prompts, new patient intake logic, emergency triage protocol, and ADA-compliant disclosure language. Workforce Wave's website crawl populated the practice-specific variables on top of that template.

The template included several dental-vertical-specific behaviors that the DSO had validated across their existing practices:

• CDT-code-aware hygiene appointment language (distinguishing prophylaxis from periodontal maintenance in scheduling conversations)
• Insurance pre-authorization guidance (specific to the 40 most common networks in the DSO's market)
• Recall reminder optimization (timing, language, and escalation paths based on recall type)
• Emergency after-hours protocol that distinguished true dental emergencies from urgent-but-schedulable situations

Each provisioned agent inherited these behaviors automatically. Practice-specific details layered on top: this practice's specific providers, their specific hours, their specific accepted networks, their specific new patient process.

The Results

Onboarding time: 50 practices provisioned in 3.5 hours, including Workforce Wave crawl time, knowledge base generation, integration setup, and phone number assignment. Previous manual process: approximately 100 hours.

Human involvement: Three practices required human review. Of those:

• One practice had no website. The DSO coordinator manually entered the practice information into the WFW dashboard (approximately 25 minutes).
• One practice had a password-protected website that Workforce Wave couldn't access. The practice manager emailed the relevant information; the coordinator built the KB manually (approximately 40 minutes).
• One practice had a website in Vietnamese, which Workforce Wave flagged for review. A bilingual team member at the DSO reviewed the Workforce Wave-generated configuration for accuracy and approved it with minor corrections (approximately 30 minutes).

47 of 50 practices: zero human involvement after the API call.

Six-month outcomes: 47 of 50 practices reported measurable reductions in missed calls. Across the cohort, missed call rate dropped from an average of 31% to 9%. Front desk time on inbound calls dropped an average of 34%.

The three edge-case practices performed comparably to the automatically provisioned ones once their manual configurations were complete — the template and intelligence layer were the same regardless of how the configuration was built.

Operations coordinator role: Transformed. She now spends her time on ongoing optimization review (Workforce Wave-generated suggestions for all 50 practices), quality monitoring, and edge case handling — rather than data entry. The DSO added two new practices in Q1; both were provisioned automatically in under four hours.

The Intelligence Loop

At the 30-day mark, Workforce Wave generated optimization reports for all 50 practices simultaneously. The DSO's VP of Technology had anticipated this would be overwhelming — 50 reports to review.

It wasn't. Workforce Wave grouped the suggestions by type and frequency. Across the 50-practice cohort, the most common optimization was a single pattern: the agent was struggling with caller questions about "new patient specials" or "new patient pricing" — promotional offers that weren't consistently listed on practice websites but were referenced in Google ads and review responses.

Workforce Wave recommended a standard addendum to the knowledge base for 34 of the 50 practices: a prompt instructing the agent to acknowledge new patient pricing inquiries and route them to a live coordinator for specifics ("I'd love to connect you with our team to go over our new patient options — let me transfer you"). The DSO approved the addendum once; it was applied across all 34 practices in under a minute.

That kind of cohort-level insight — identifying a pattern across 50 locations and suggesting a unified fix — wasn't available when each practice was configured individually. The scale created analytical leverage.

What They'd Tell You

The VP of Technology, at the six-month mark:

"We had always assumed voice AI would require a manual setup call for each practice. The three edge cases were the first time we looked at the system in a month. The other 47 just worked. What I didn't expect was the cohort intelligence — when Workforce Wave surfaces the same issue across 30 practices at once, you can fix it once and roll it out everywhere. That's not possible when each configuration is hand-built. We're now looking at using Mode 3 to provision agents for referral partners who aren't even in our DSO. The API just works."