OpenAI has announced a partnership with Retell AI to offer customizable voice agent automation powered by GPT-4o [source]. The integration allows enterprises to deploy conversational AI agents without writing code, using a visual interface to configure call flows, prompts, and voice parameters.

According to the announcement, Retell AI's platform connects to OpenAI's Realtime API, enabling low-latency voice interactions for customer service, appointment scheduling, and sales qualification. Users can select from multiple voice profiles and adjust response styles through dropdown menus and text fields.

The partnership follows a pattern of OpenAI embedding its models into third-party automation platforms. Retell AI joins a roster of voice AI providers—including Vapi, Bland AI, and Air AI—that have integrated GPT-4o for telephony applications since the model's release in May 2024.

OpenAI's blog post highlights use cases in healthcare, hospitality, and logistics, where voice agents handle inbound inquiries and outbound campaigns. The company states that Retell AI customers have processed "millions of calls" using the integration, though no independent verification of call volume or accuracy rates is provided.

The announcement does not disclose whether OpenAI conducted internal testing of Retell AI's implementation, nor does it specify guardrails for preventing hallucinations in voice outputs. Voice-based large language models have previously been observed generating plausible but incorrect information in customer-facing scenarios, a risk amplified when users cannot easily verify claims in real time.

No pricing details or service-level agreements were included in the announcement. The integration is available immediately to Retell AI customers with OpenAI API access.