For Wavenote, the value of a recording continues long after a conversation ends. Novarcs helped the company bring its core AI services onto AWS, making it easier for users to turn recorded discussions into useful summaries, clear action items and searchable knowledge.
A product built around recorded conversations
Wavenote combines recording hardware with mobile and web applications for professionals and teams. Its users capture meetings, interviews and other spoken information, then use AI to transcribe the recording, find the important points and return to the content when they need an answer.
The business brings together hardware sales, AI subscriptions and transcription allowances. That makes the quality of the cloud service part of the everyday product experience. A useful recording needs a dependable transcript, a readable summary and a practical way to recover the details that matter.
When AI consistency became a product issue
As the service grew, Wavenote was managing several model services with different operating and billing arrangements. Summaries could vary in format and style, making professional templates less predictable and creating more work to check the output. Sensitive meeting content also needed a clearer approach to handling and access.
AI spending was another concern. Usage varied across subscribers and features, but the company could not easily connect that consumption to the service being delivered. Wavenote needed to improve the user experience while giving its product and operations teams a clearer view of how the service was running.
A consistent workflow from transcript to answer
Novarcs migrated the core AI workflows to AWS, using Amazon Bedrock for the generative AI capabilities and Amazon Transcribe for speech transcription. The delivery focused on the functions users already relied on: meeting summaries, action-item extraction, questions over meeting content and professional templates.
The team reworked prompts and templates for those tasks and added checks around the structure of generated output. This gave summaries and follow-up information a more consistent shape, while keeping questions grounded in Wavenote's own meeting and document content.
Moving the service into everyday operation
Novarcs carried out the application and data transition in stages, validating the core workflows before completing the production move. The work included operational monitoring, support documentation and handover, so Wavenote could continue running and improving the service after delivery.
The project also established AI usage and cost attribution by user and feature. Wavenote gained a clearer basis for understanding the cost of its subscription services and deciding where further optimization would be useful, alongside a simpler way to manage its AI services.
Meeting knowledge that users can put to work
With the core workflows running on AWS, users could work with more stable transcription, structured meeting summaries and clearly captured action items. Questions about a recording became a practical way to recover relevant information, helping people return to the substance of a discussion when they needed it.
User feedback described clearer summaries, more dependable recognition in mixed-language conversations and a useful day-to-day question-and-answer experience. For Wavenote, the completed project connected those product improvements with better visibility into AI operations and cost, supporting an ongoing service that its team could manage with greater confidence.