## Summary so far
- Introductions underway among participants.
- Colin Heilbut from Rime is facilitating.
- Participants include Nick, director of conversation AI, Sarah (leads engineering/contact center), and Mike (contact center).
- Role-play setup: Colin as AE from Rime selling to three people from Assurant.
- Roles clarified: Nick (director of conversation AI), Sarah (engineering/contact center), Mike (contact center).
- Brief agenda shared by Colin: overview of Rime, differentiation from other TPS, focus on Voice AI in insurance.
- Colin prepared a brief live demo relevant to participants' work.
- Discussion to address concerns about voice quality and other challenges.
- Colin introduced a digital whiteboard tool to capture success criteria and notes during the meeting.
- Initial focus on voice quality issues and their impact on metrics like containment and NCSAT.
- Voice quality problems cause poor task completion.
- Success criteria discussed around improving CSAT.
- Sentiment analysis tools currently used to measure customer sentiment.
- Emphasis on tracking success criteria to ensure meaningful discussion.
- Additional voice AI challenges identified: latency, pronunciation of brand and product names, handling quick interruptions.
- Severity and measurement of issues like pronunciation and latency discussed.
- Colin to demonstrate how Rime typically addresses these challenges.
- Ease of deployment discussed; Rime compatible out of the box with NICE, can be deployed via toggle or docker file, enabling same-day setup.
- Colin shared Rime’s founding in 2022 by Lilly, Brooke, and Ari.
- Rime built multiple recording studios worldwide to produce higher quality training data with real speakers.
- Rime powers millions of phone calls monthly for major restaurant chains and healthcare groups.
- Rime sponsored a 100,000 call survey showing significantly lower hang-up rates and fastest time to talk compared to other AI.
- Data from a large telecommunications provider showed a 25% increase in call containment using Rime.
- Additional data showed a 20% increase in food sales from a national food brand.
- Colin began a live demo simulating a customer call for an extended warranty claim.
- Demo scenario: Colin as customer reporting broken phone, verifying identity, claim details, and requesting specialist transfer.
- Demo highlights Rime’s conversational capabilities, natural flow, and claim processing steps.
- Demo included identity verification, policy check, claim initiation, secure upload link offer, and specialist handoff.
- Demo ended with a brief French language interaction.
- Colin asked for feedback on latency and overall call experience.
- Colin explained voice selection process from over a thousand voices, focusing on performance and user response, with a tool for clients to pick voices.
- Discussion on pilot decision-making process, involving other teams like security and contract considerations.
- Colin recapped key pain points: latency, pronunciation, voice quality affecting containment and CSAT.
- Rime claims to have some of the lowest latency rates in the industry with products Mist and Coda.
- Pronunciation customization available, with Koda offering out-of-the-box pronunciation and Mist allowing more tailored setups.
- Colin noted ongoing interest in speech-to-speech technology, though no public announcements yet.
- Meeting concluded with a brief explanation of how Rime generally highlights key points.

## Key points
- Colin Heilbut is the founding AE at Rime.
- Nick leads go-to-market and is director of conversation AI.
- Sarah leads engineering/contact center.
- Mike is from the contact center.
- Role-play exercise planned with Colin selling to Assurant representatives.
- Rime differentiates itself from other TPS providers.
- Focus on Voice AI applications in insurance.
- Concerns about voice quality and other challenges acknowledged.
- Digital whiteboard used to capture discussion points and success criteria.
- Voice quality problems impact containment, NCSAT, and task completion.
- CSAT improvement is a key success metric.
- Sentiment analysis tools are currently in use to measure customer sentiment.
- Tracking success criteria is critical for effective evaluation.
- Additional voice AI issues include latency, brand/product name pronunciation, and managing quick interruptions.
- Severity of pronunciation and latency issues assessed to prioritize solutions.
- Rime offers easy deployment compatible with NICE systems and orchestration stacks, enabling rapid setup.
- Rime’s unique approach includes proprietary recording studios for high-quality training data.
- Rime powers millions of calls monthly across various industries.
- Independent survey data supports Rime’s superior customer engagement metrics.
- Case study data shows 25% call containment improvement for a major telecom client.
- Case study data shows 20% increase in food sales for a national food brand.
- Live demo simulated a customer interaction for extended warranty verification.
- Demo showed natural conversational flow, identity verification, claim initiation, secure upload link, and specialist handoff.
- Demo included multilingual support (French).
- Colin solicited feedback on latency and voice quality after demo.
- Voice selection process involves rigorous testing of over 1,000 voices to find best performing and most well-received options.
- Clients can select voices that best meet their needs via a provided tool.
- Pilot decision-making involves multiple stakeholders including security and contract teams.
- Rime’s products Mist and Coda offer industry-leading low latency and voice quality.
- Koda provides out-of-the-box pronunciation accuracy; Mist allows customized pronunciation setups.
- Rime is exploring speech-to-speech technology for future development.

## Decisions
- None yet.

## Action items
- Colin to run the role-play exercise with Assurant team.
- Colin to deliver a brief live demo focused on Voice AI relevant to insurance and participants' work.
- Colin to track and capture success criteria during discussion to ensure measurable outcomes.
- Colin to demonstrate how Rime typically helps address voice quality, latency, pronunciation, and interruption handling.
- Colin to explain Rime’s deployment options and compatibility with existing contact center infrastructure.
- Colin to gather participant feedback on demo performance, including latency and voice quality.
- Colin to discuss pilot decision-making process and identify other teams to involve (e.g., security, contracts).

## Open questions
- What specific voice quality problems are causing issues?
- How do voice quality issues affect containment and NCSAT?
- What constitutes a significant CSAT improvement for the participants?
- How should latency and pronunciation issues be addressed in the solution?
- What capabilities are needed to handle quick interruptions effectively?
- Are there any additional pain points or challenges to consider before the demo?
- How did participants perceive the latency and overall voice quality in the demo?
- What is the expected timeline and criteria for pilot success?
- Who else needs to sign off on a project like this within the participants’ organization?