Angle Health Handed 80% of Provider Calls to AI Agents, Which Resolved 90% Without Human Handoff

A call from a clinic sounds simple: confirm whether a patient's insurance covers a procedure. But getting an answer often means spending long minutes on hold. For an insurer, that adds up to thousands of repetitive calls every day, consuming the time of staff needed for genuinely complex cases.

Angle Health, which administers health plans for small and midsize businesses, decided to hand this routine work to voice agents on the ElevenAgents platform. This is a case study from the vendor itself, so its figures should be read with marketing caveats. Still, the architecture of the solution is noteworthy.

What the agents actually do

About 80% of incoming calls to the provider operations team come from clinics and concern eligibility and benefits checks, prior-authorization status, or claim status. This is the traffic the agents now handle.

The pilot is built around two agents with clearly defined roles. The first verifies the caller's identity for compliance, retrieves live claims and benefits data, and submits prior-authorization requests. The second answers questions about eligibility, benefits, claim statuses, and member ID numbers. Both connect directly to the administration platform's core system: rather than reciting a knowledge base, they read real data and take actions.

Another detail: once an escalated claim has been resolved, the agent calls the clinic and explains the outcome. This is no longer reactive support but an outbound process.

Pilot results

The agents handled about 80% of total provider call volume, and more than 90% of those calls were resolved end to end without a human handoff. The team reclaimed nearly 9,000 minutes a day. Responses are available around the clock in eight languages. According to the company's engineers, when daily call volume through the platform increased tenfold, it scaled without issues.

Angle's product lead separately highlighted the remaining 10%: the agent classifies why it could not resolve the issue and routes the call to the appropriate team. For the product, this matters more than an attractive automation percentage: a high-quality handoff reduces the cost of errors.

Healthcare, compliance, and vendor comparisons

The stakes are higher in healthcare. Angle deploys the agents only under a BAA and with HIPAA safeguards, including Zero Retention Mode, in which the provider does not retain conversation data. Clinical terminology and diagnosis codes must be captured accurately.

According to Angle, ElevenLabs was tested alongside several other vendors and ranked best for transcription accuracy and task execution in a healthcare context. The company did not disclose its comparison methodology, so the claim cannot be independently verified.

Production uses deterministic workflows that log required information and follow rules for human handoff. Built-in call analysis allows the team to verify that the rules are being followed.

Limitations and takeaways

This is a pilot covering selected scenarios, not a complete replacement for a call center. There is no data on error rates, clinic complaints, per-call costs, or how "resolved end to end" was defined. The source is the vendor's blog, and the quotes come from a customer with an interest in continuing the partnership.

The practical lessons for teams building voice agents in regulated industries are:

• Choose a narrow workflow with recurring questions and clearly structured backend data.
• Give the agent access to live systems, not just documentation.
• Build in caller verification and deterministic rules for human handoff.
• Measure not only the automation rate but also the quality of triage for the remaining cases.
• Compare vendors using your own call recordings, not demos.

Angle plans to expand the pilot into commercial operations, broker services, and account management. If the results hold at scale, it will provide a useful benchmark for voice agents moving beyond the experimental stage in insurance.