Voice agent for abandoned intake recovery
Tryozi's voice agent gives unfinished intake a clear route back.
An AI voice agent helps eligible, opted-in users resume an incomplete intake, handling administrative questions and passing anything clinical to the right team.
The impact
29%
of contacted users resume intake
Voice agent for abandoned intake recovery
Case study facts
- Industry
- Telehealth
- Team
- Marketing
- Systems connected
- 4
- Time to live
- 3 weeks
- Engagement
- AI implementation
The impact
Results for the team
- of contacted users resume intake
- 29%
- of contacted users resume intake
- The recovery rate counts contacted users who return to their incomplete intake.
- less manual follow-up time
- 62%
- less manual follow-up time
- The voice agent handles routine administrative follow-ups, reducing the time staff spend contacting users manually.
- more eligible follow-ups covered
- 3×
- more eligible follow-ups covered
- The team covers more eligible, opted-in users with recovery follow-ups, using clear paths for resuming intake or requesting support.
Situation
Intake stopped. The reason was unclear.
An unfinished intake does not explain why someone stopped. Generic reminders cannot tell a technical problem from a question that needs a person, and manual follow-up competes with other support work.
Challenge
A reminder could not resolve every problem.
Follow-up needed to tell an administrative obstacle from a question requiring a person. Contact permission, identity, opt-outs and clinical boundaries had to remain explicit throughout the conversation.
What we built
A guided path back to intake.
The workflow identifies incomplete intake records eligible for follow-up and checks contact permission before every call. The voice agent identifies itself, asks whether help is wanted and guides the user back to a secure continuation path.
Clinical questions, uncertainty or requests for a person move to the appropriate team; the call is never a shortcut around the intake process.
- 01
Detect unfinished intake
Find incomplete records without assuming why the user left.
- 02
Verify contact permission
Consent, suppression and contact-frequency rules are enforced before every call.
- 03
Load minimal context
The agent sees only the administrative details the call requires.
- 04
Offer help
- Voice agent
The voice agent discloses that it is automated and confirms the user wants to continue.
- 05
Resume or hand off
- Human in the loop
Send a secure continuation link or connect the user to a person.
- 06
Log the outcome
Record the disposition and honor opt-out or follow-up preferences.
Evals and guardrails
Test conversations. Enforce contact boundaries.
Conversation evals
Test whether the agent understands requests and completes intake support tasks. Include interruptions, voicemail, wrong recipients and opt-outs.
Handoff evals
Measure missed clinical questions and requests for a person. Test failed transfers and check that the agent never gives clinical advice.
Guardrails
Check contact permission before each call. Honor opt-outs, limit access to intake details and send clinical questions to qualified staff.
Results
Unfinished journeys have a next step.
The recovery rate counts contacted users who return to their incomplete intake. The voice agent handles routine administrative follow-ups, reducing the time staff spend contacting users manually.
The team covers more eligible, opted-in users with recovery follow-ups, using clear paths for resuming intake or requesting support.
- of contacted users resume intake
- 29%of contacted users resume intake
- less manual follow-up time
- 62%less manual follow-up time
- more eligible follow-ups covered
- 3×more eligible follow-ups covered
From Tryozi
“We had tried other partners and nobody could unblock it. Revensi came in, understood the problem quickly, and turned a bottleneck into something that just runs. Total game changer.”
Build Capacity To Grow. Own Your Intelligence.
Bring the workflow slowing your sales, orders or delivery. We'll map the bottleneck and show you what the AI system could look like.