CUSTOMER CASE
How Confido Health is Safely Scaling AI Voice Agents Across Healthcare Operations
We needed to test complex multi-node workflows and backend integrations with confidence - ensuring every patient interaction works flawlessly before going live.
Co-founder & CPO at Confido Health
Company
About
Confido Health empowers specialty healthcare providers by deploying AI "digital workers" that automate administrative workflows end-to-end - from appointment scheduling to insurance verification and care coordination. Growing at blitz speed, Confido is revolutionizing healthcare operations with intelligent voice agents that handle complex, multi-turn conversations.
Industry
Healthcare Technology
Company size
11-50
Key features
- • AI voice agents
- • Appointment scheduling
- • Insurance verification
- • Care coordination
- • EHR integration
The Challenge
Confido's ambitions to scale and modernize its voice infrastructure encountered four major obstacles:
1. Complex Multi-Node Workflows
Confido's voice agents do more than simple menu navigation - they drive branching logic across multiple nodes. For example:
- Node A: verify patient identity (multi-turn prompt)
- Node B: determine workflow type (appointment, refill, insurance)
- Node C: follow-up confirmation or fallback
Each node has a unique prompt and state logic, and transitions depend on upstream responses. Manually testing all permutations quickly became unwieldy.
2. Tool-Call Testing (API / Backend Integrations)
Because Confido's agents don't just talk - they invoke downstream tools (EHR lookups, insurance verification APIs, text/SMS systems, CRM updates) - it was critical to test not only conversational logic but also that tool calls worked correctly under various failure modes (timeouts, partial data, retries). Without automation, every edge case would require painstaking manual verification.
3. Infrastructure Migration Risk
Confido was transitioning its entire voice call infrastructure - effectively migrating its live voice routing, telephony integration, and agent endpoints. The risk: any performance degradation, increased latency, dropped calls, misrouted prompts, or tool-call failures during or after cutover would directly hurt patient experience. Without rigorous testing, the migration felt risky.
The Solution
By embedding Cekura into its QA pipeline, Confido addressed each challenge:
Automated Scenario Generation & Multi-Node Testing
Before pushing changes, Confido automatically generated test scenarios for each workflow variant (based on prompt definitions and branching logic). The system would spin up parallel simulations, walking each node, verifying state transitions, verifying fallback behaviors, and cross-checking expected downstream tool call outputs.
"With one click, we simulated our 30+ service workflows - scanning every node and edge - saving days of manual QA."
This shift allowed Confido to ship new workflows, updates, or prompt tweaks across dozens of clinics with confidence.
Tool-Call & Backend Verification
Each test run not only validated speech-level flows but also ensured that tool calls executed correctly under real-world conditions. For instance, Cekura could simulate a patient requesting an appointment for next week and verify that Confido's agent correctly invoked the scheduling tools to fetch available slots, then dictated those options back to the simulated patient. This guaranteed true end-to-end accuracy - validating both conversation logic and backend execution.
Migration Stress-Testing & Canary Runs
Ahead of migrating the calling infrastructure, Confido ran full-scale stress tests - simulating thousands of calls. They validated:
- Call latency
- Workflow accuracy
- Tool-call success rates
"We set up key metrics on Cekura and could easily compare it between our old and new stack. This gave us the confidence to deploy the new stack."
The Results
By coupling Confido's voice agents with robust automated assurance and monitoring, the migration and scaling succeeded with minimal risk - and with measurable gains:
- Zero regression on key workflows - every node and integration preserved post-migration
- Increased deployment velocity - new services rolled out in hours instead of days
- Reduced manual QA burden, allowing engineers to focus on improvements, not regression tests
Confido's engineering and operations teams have set a new benchmark for reliability in healthcare automation. Their commitment to precision, patient experience, and continuous improvement shines through every voice interaction. We are proud to be partners in their journey.