Jobcase Delivers Faster, More Natural Voice Experiences with Deepgram Aura-2

Jobcase helps millions of members find jobs and move forward in their careers, with a focus on timely, human-centered support for frontline and hourly workers. As the company expanded its use of AI assistants and automated outreach, it needed a way to make those interactions feel as intuitive and trustworthy as talking to a real person.

Deepgram’s Aura-2 text-to-speech (TTS) capabilities now power Jobcase’s AI-driven voice experiences, reducing latency, improving scalability, and delivering a more fluent first impression on every call.

• More fluent, natural calls for job seekers, especially in the crucial first few seconds of an interaction.
• Reduced TTS latency and greater scalability, enabling higher call throughput without sacrificing quality.
• Best overall tradeoff of quality, cost, and initial latency versus other leading TTS providers Jobcase evaluated.
• On‑prem deployment path validated, giving Jobcase confidence it can scale concurrency and manage performance as volumes grow.

Jobcase’s mission is to help its members find jobs and progress in their careers—often in moments where time, clarity, and support matter most. As the team looked for ways to expand outreach and support without overloading human agents, voice agents emerged as a powerful solution.

Voice agents promised:

• Personalized communication at scale, allowing Jobcase to speak with more members without losing a sense of connection.
• Always-on availability for job seekers who might be juggling multiple jobs, irregular shifts, or limited access to desktop devices.

Through testing and live deployments, Jobcase found that latency and voice quality directly influenced member engagement, particularly at the very start of a call. If the voice sounded robotic or there was an awkward delay before it spoke, members were more likely to disengage.

Lesson learned: With AI voice agents, first impressions matter—both in how quickly the system responds and how natural it sounds.

Jobcase uses Deepgram primarily as a text-to-speech provider for AI-driven inbound and outbound calls, with speech-to-text integrated into the same pipeline for full conversational flow.

• Member speaks: The member speaks during an inbound or outbound AI-driven call.
• Audio capture and transcription: Once the member stops speaking, that audio segment is captured and sent to Deepgram for speech-to-text transcription.
• AI understanding and decisioning: The transcribed text is processed by an AI-driven handler, which interprets the member’s intent and determines the next best response.
• Response generation and synthesis: The AI generates a text response, which is then sent to Deepgram Aura-2 for text-to-speech conversion.
• Natural voice playback: Deepgram returns the synthesized audio, which is played back to the member in real time, continuing the conversation in a natural, low-latency loop.

This architecture lets Jobcase tightly control latency and call flow while keeping its vendor stack flexible and configurable.

During evaluation, Jobcase tested and integrated multiple TTS providers, including AWS Polly, OpenAI, and ElevenLabs. Across these comparisons, Deepgram Aura-2 stood out for delivering the best overall tradeoff between:

• Voice quality: Natural, clear delivery suited for production calls.
• Cost: Competitive pricing that made it feasible to run large-scale AI voice workloads.
• Initial latency: Fast time-to-first-audio that helped avoid the awkward pauses that cause members to drop off.

The team prioritized:

• Stability and reliability: The system needs to perform consistently in production, not just in demos.
• On‑prem deployment option: Aura-2’s on-premise path for scalable concurrency and managed performance was a key selling point, ensuring Jobcase can maintain control as it grows.

Importantly, Jobcase’s system is designed for heterogeneous vendor integrations. Having both TTS and ASR from a single provider is not a hard requirement, and the platform can mix and match providers as needed.

Deepgram earned its place in this environment on performance and reliability, not vendor lock-in.

Since deploying Aura-2, Jobcase has seen meaningful improvements in both technical performance and end-user experience, ensuring faster, smoother calls. Deepgram has provided reduced TTS latency and greater scalability, which directly contributes to:

• Increased call throughput – More calls can be handled in parallel without degrading the experience.
• A more fluent, conversational user experience, with less dead air and more natural pacing.

Internal stakeholders at Jobcase have anecdotally observed a clear improvement in the user experience when using the Aura-2 model, particularly in how quickly and naturally the system responds to members.

While Jobcase is not sharing public call-volume statistics at this time, these qualitative signals have reinforced the team’s confidence that Deepgram is improving the overall quality of their AI-powered interactions.

Jobcase has validated the suitability of Deepgram’s on‑prem solution for its use cases, and sees on‑prem deployment as a likely next step as volumes continue to grow.

Priorities going forward include scaling AI voice interactions to reach more members without sacrificing quality, leveraging on‑prem for tighter control over concurrency, performance, and potential future cost optimizations, and continuing to refine vendor mix and pipeline configuration to match evolving performance and business requirements.

Beyond the core technology, Jobcase has found Deepgram’s customer support and engagement to be strong, with accurate, comprehensive documentation that made integration and ongoing maintenance more straightforward.

As Jobcase continues to build out conversational AI in the employment sector, Deepgram remains a key part of its stack, enabling the company to deliver faster, more natural voice experiences that help members move forward in their careers at scale.