# Meeting summary

Assurant, a leading provider of warranty services with major clients like T-Mobile and Geico, is evaluating Rhyme's conversational Voice AI technology as a potential upgrade from their current Microsoft-based TTS solution (via NICE's Cotton Energy). The Assurant team, including the Chief Product Officer, head of customer-facing voice, engineering lead, and contact center operations manager, is focused on improving customer experience, reducing human agent workload, and addressing voice quality issues. The main outcome was agreement to set up a dev/test environment for a pilot, with a collaborative approach to defining success criteria and a decision timeline of 3-4 months.

## Pain points & concerns

- **Voice quality (pronunciation, expressiveness, prosody, latency):** Customer complaints have been received about the current Microsoft-based TTS voices, impacting customer satisfaction and brand perception. Severity: Moderate to high, as it is the first customer touchpoint.
- **Static, deterministic IVR experience:** Current system is limited in flexibility and dynamic interaction, restricting use cases and customer engagement. Severity: Moderate, as Assurant is early in their voice AI journey but recognizes the need to evolve.
- **Maintaining accuracy and control:** Any new solution must match or exceed current pronunciation accuracy, and the team is cautious about losing manual control as they move to more agentic, dynamic AI. Severity: High, as accuracy and reliability are non-negotiable.
- **Operational efficiency and capacity:** Need to maximize AI containment to reduce human agent workload and avoid increased staffing or customer hold times. Severity: High, as it directly impacts costs and service levels.
- **Compliance and deployment flexibility:** Assurant operates in Microsoft Azure data centers and requires options for compliant, secure deployment (cloud or on-premises). Severity: High, due to regulatory and client requirements.
- **Organizational changes:** Ongoing layoffs may affect team capacity and increase pressure to automate. Severity: Moderate, with potential impact on project urgency and resourcing.

## Success criteria for the new technology

- **Voice quality:** Natural, expressive, and accurate pronunciation, with the ability to teach new words once for consistent replication.
- **Task completion and call containment:** Maximize the amount of information gathered by AI before human handoff; increase call containment rates.
- **Latency:** Fast, responsive voice interactions, competitive with or better than current benchmarks.
- **Brand and name pronunciation:** Precise, consistent handling of brand-specific terminology.
- **Integration and deployment:** Seamless integration with NICE and Microsoft Azure environments; support for SOC 2 Type 2 compliance.
- **Language support:** Support for at least English and French, with openness to additional languages as needed.
- **Pilot evaluation:** Success metrics to be collaboratively scoped, including call completion, brand pronunciation, latency, and customer experience as measured in the dev/test environment.

## Decision process & criteria

- **Stakeholders:** Chief Product Officer, head of customer-facing voice, engineering lead (Patrick), and contact center operations manager (Walt) are all involved; Patrick and his team will lead technical evaluation and pilot setup.
- **Champion:** Patrick (engineering lead) is the primary technical contact and likely champion.
- **Evaluation approach:** Phased pilot starting in a constrained dev/test environment, with baseline security checks and documentation review. Success criteria to be defined collaboratively.
- **Approval process:** Requires baseline security review, documentation, and possibly completion of a GenAI form; authority exists within the team to approve dev/test pilots.
- **Paper process:** Not fully detailed, but involves standard security and compliance checks before production rollout.

## Timeline

- **Pilot setup:** Immediate next step is to set up a dev/test environment for initial evaluation.
- **Decision window:** 3-4 months for evaluation and decision-making.
- **Phased approach:** Start with dev/test, then scope technical requirements for on-premises or production pilot.
- **Urgency signals:** Organizational changes (layoffs) and desire to reduce human agent workload may add urgency.

## Next steps

- [ ] Set up Rhyme voice AI in Assurant's dev/test environment — Patrick (Assurant) & Colin (Rhyme)
- [ ] Complete baseline security checks and provide required documentation — Colin (Rhyme) to supply, Patrick (Assurant) to review
- [ ] Identify key voices and use cases for initial demo and pilot — Assurant team with guidance from Rhyme
- [ ] Schedule follow-up scoping call (next Wednesday or Thursday) — Colin (Rhyme) & Assurant team
- [ ] Define pilot success criteria and evaluation metrics collaboratively — Both teams
- [ ] Explore technical requirements for on-premises deployment if pilot is successful — Patrick (Assurant) & Rhyme technical team
- [ ] Provide direct contact info for ongoing support — Colin (Rhyme)
- [ ] Prepare for completion of GenAI form and any additional compliance documentation — Assurant team