Retell Expands Into Contact Center OS

This move says Retell is trying to graduate from being a voice API that developers wire together into being the operating system for AI support. Instead of only handling the live phone conversation, it is pushing into the layers around the call, deciding which agent gets the call, checking whether the agent followed policy, and running test calls before launch. That broadens Retell from a usage fee on minutes to a larger share of contact center software spend.

• Retell started as a developer first voice layer, where teams write prompts, connect tools, pick speech and LLM providers, and pay a $0.055 per minute platform fee on top of model costs. Building routing, QA, and testing on top of that gives it more of the workflow that sits around every production deployment.
  1 sacra 2 sacra
• The comparison set splits into two camps. Vapi is closest to Retell as a flexible builder for developers, while Sierra and Decagon sell more complete support systems, often with high touch implementation, outcome based pricing near $1.50 per resolution, and responsibility for customer experience metrics like containment and CSAT.
  1 sacra 3 sacra 4 sacra 5 sacra
• That matters because the real product in AI support is not just a talking bot. Buyers want the full loop, ingest docs, route the conversation, take actions in CRM and billing tools, measure whether the answer was correct, and improve performance without hand auditing thousands of calls. Product surface, not just model quality, is where differentiation has shifted.
  3 sacra 4 sacra 5 sacra

The next phase of the market is a collision between developer controlled voice infrastructure and full stack AI support suites. If Retell keeps moving up from call handling into management software, it can win larger contracts and become harder to rip out, while pressure rises on Vapi from below and on Sierra and Decagon from a faster moving self serve alternative.