Call center ai · Production

Super cuts STT costs by ~30% and scales real-time voice agents for real estate with AssemblyAI Universal-3 Pro Streaming

The problem

Super's real-time phone agents were limited by STT providers that offered only basic turn detection timing controls, hit concurrency ceilings at scale, and provided keyterm prompting with minimal real-world impact for real estate vocabulary.

First attempt

Multiple prior STT providers were tested and consistently fell short on turn detection, concurrency, and keyterm prompting, forcing workarounds that consumed engineering resources.

Workflow diagram · grounded in source
1
Phone call reaches voice agent
trigger
“real-time phone agents that field thousands of calls daily across inbound and outbound use cases”
2
Real-time speech transcription
ai_action
“Universal-3 Pro Streaming is engineered for real-time voice applications, with improvements to turn detection, multilingual handling, and keyterm prompting that meaningfully boosts the terms developers prioritize, rather than functioning…”
3
Multi-agent orchestration
ai_action
“Super's multi-agent architecture allows for hyper-specialized agent configurations, and the availability of mid-stream parameter updates supports fine-tuning for specific agent use cases without sacrificing real-time performance”
4
Voice agent responds to caller
output
“End callers experience clearer conversations with more natural turn-taking, while Super's customers receive higher-quality transcription outputs that lead to better resolutions for their own end customers”
Reported outcome

Universal-3 Pro Streaming delivered approximately 30% cost savings on STT, freed the engineering team from reactive maintenance to focus on product development, and improved conversational quality across thousands of daily calls.

Reported metrics
STT cost savings~30%
Integration timeapproximately one day
Engineers required for integrationone engineer plus QA support
Daily call volumethousands of calls daily
Show all 5 reported metrics
STT cost savings~30%
integration timeapproximately one day
engineers required for integrationone engineer plus QA support
daily call volumethousands of calls daily
engineering capacity reallocationreallocated capacity toward new feature development
Reported stack
AssemblyAIUniversal-3 Pro StreamingLiveKit
Source
https://www.assemblyai.com/customers/super
Read source ↗

Frequently asked questions

What did this team achieve with this AI workflow?

Universal-3 Pro Streaming delivered approximately 30% cost savings on STT, freed the engineering team from reactive maintenance to focus on product development, and improved conversational quality across thousands of…

What tools did this team use?

AssemblyAI, Universal-3 Pro Streaming, LiveKit.

What results were reported?

STT cost savings: ~30%; Integration time: approximately one day; Engineers required for integration: one engineer plus QA support; Daily call volume: thousands of calls daily (source-reported, not independently verified).

What failed first in this deployment?

Multiple prior STT providers were tested and consistently fell short on turn detection, concurrency, and keyterm prompting, forcing workarounds that consumed engineering resources.

How is this call center ai AI workflow structured?

Phone call reaches voice agent → Real-time speech transcription → Multi-agent orchestration → Voice agent responds to caller.

Related call center ai cases