The Challenge: Voice AI to Handle Complex Customer Service Challenges

Lorikeet is an AI concierge for complex businesses that understands customer context and takes action on their behalf, 24/7, across any channel. The Concierge Agent helps companies deliver exceptional customer support by actually resolving their issues in-session, rather than linking them to an FAQ or defaulting to a human. Lorikeet was co-founded by Jamie Hall, formerly of Google Brain where he helped build one of the early foundational language models, and Steve Hind, formerly of Stripe and Watershed.  

For Lorikeet's customers in regulated industries like financial services, healthcare, and insurance, the stakes on every call are high. A cardholder's payment just got declined at the grocery store. A patient's medication hasn't arrived. These are moments of real stress, and the voice on the other end of the line needs to build trust instantly—guiding the caller through authentication, account lookups, and multi-step resolutions while staying within compliance guardrails.

Lorikeet needed a text-to-speech platform that could keep pace with this complexity: natural-sounding voices with low enough latency that conversations flow without awkward pauses, and infrastructure reliable enough to handle sudden volume spikes without degradation.  

"We put a lot of R&D effort into getting our voice agents to a level where they could actually help people with complicated or demanding customer situations, but with a latency profile that means the conversation sounds natural and the customer is not left hanging on the phone," said Jamie Hall, Lorikeet's co-founder and CTO. "That's only possible if you have very fast text-to-speech technology, which Cartesia is particularly excellent at."

Why Lorikeet Chose Cartesia

Lorikeet evaluated multiple voice platforms against four criteria they needed: low latency for fast and realistic response times, reliability at scale to handle concurrent calls, voice quality to adapt to human speech patterns, and enterprise partnership. Cartesia won decisively across all fronts.

Speed was a clear differentiator: Cartesia was almost half a second faster than competing vendors. In live support calls, that margin is the difference between a conversation that flows naturally and one that triggers an escalation request. 

Cartesia's curated voice library also proved more reliable for production deployment than vendors with larger catalogs that required extensive testing before release. Further, Cartesia provided a strong partnership relationship with responsive support, including in-person meetings at Lorikeet’s Sydney office for product feedback. 

"Being able to easily ask Cartesia to 10x our volume in a day is very important," Hall said. "Whenever we’re under a lot of time pressure, Cartesia really steps up. I would describe the team as very responsive and great to work with."

How Cartesia Powers Lorikeet's Voice Agents

Building Trust Through Natural Conversation

Lorikeet uses Cartesia voices in its agents to build trust through natural pacing, emotional balance, and carefully tuned disfluencies—the small "ums" and pauses that make a conversation feel human rather than robotic.

The result: callers regularly don't realize they're speaking with AI, even though Lorikeet discloses it in calls. Customers thank the agent and tell it to have a nice day. The voice experience earns enough trust for callers to stay engaged through complex, multi-step resolutions rather than demanding a human agent.

"We’re appreciative of Cartesia’s ability to have emotional balance in the voices, because empathy and an emotional response is very important for a phone support interaction to go well,” said Hall.

Resolving Real Crises at Scale: GiveCard Deployment

When 300,000+ families had their SNAP food benefits suddenly transitioned to GiveCard, Lorikeet deployed voice agents within 24 hours to handle the surge. Callers were stressed and workflows required navigating complex paths through authentication and troubleshooting in English, Spanish, and Mandarin.

Powered by Cartesia, Lorikeet handled over 21,000 calls in the first nine days alone, with hundreds of concurrent calls at peak, maintaining 100% uptime with no degradation issues.

“If you imagine being in that situation, talking to a very robotic voice or feeling like you're on an automated phone tree would add to the stress and anxiety. Whereas responding naturally and communicating helps put you at ease and focus on the agent being able to resolve your issue instead of deflect you,” said Hall.

Enterprise-Grade Architecture for Regulated Industries

Lorikeet serves customers across financial services, healthcare, insurance, and travel — industries where errors are costly or legally problematic. They partner with Cartesia to maintain guardrails and compliance checks at every step, including for SOC 2, HIPAA, GDPR, and other compliance standards. 

Results: Voice AI That Drives Resolution and Customer Trust

With Cartesia, Lorikeet delivers voice experiences that earn customer trust and drive resolution across regulated industries:

• Handling hundreds of thousands of calls per month across enterprise customers including Airwallex, Taptap Send, and GiveCard

• Achieved results for customers including 97% faster resolutions and nearly 2x CSAT scores

• Deployed voice agents within 24 hours for GiveCard, handling 21,000+ calls in nine days with hundreds of concurrency and 100% uptime

• Cartesia TTS response latency almost half a second faster than competing providers, keeping conversations natural and reducing escalation to human agents

• Callers regularly not realizing they're speaking with AI — engaging naturally, completing multi-step resolutions, and thanking the agent at the end of the call

Looking Ahead

As Lorikeet scales rapidly, Cartesia remains core to their voice infrastructure. They advise other enterprises to choose vendors who are invested in their success, citing Cartesia’s responsiveness to feature requests, easy access to a shared Slack, and in-person product feedback sessions. The roadmap alignment is clear: faster, more natural text-to-speech at production scale.

"Cartesia has fast and reliable text-to-speech that can handle very large-scale deployments," Hall said. "Continuing to be more natural sounding with lower latency at production scale is exactly where our product is heading and where we believe the market is heading."