Delivering on commitments made at its April 2026 launch, the Singapore-hosted transcription API marks the first production release of Toku’s open-source conversational AI initiative, with additional capabilities planned through 2026.
SINGAPORE, July 1, 2026 /PRNewswire/ — Toku, Asia Pacific’s leading AI-powered 360° customer experience platform, today made Makimoto Kawa (“Kawa”) available to developers, the first open-source conversational AI infrastructure built and hosted in Singapore, designed for teams operating under APAC’s regulatory and data residency requirements. Kawa’s post-conversation transcription API, together with an interactive playground, is live now at makimoto.ai.
The release is Toku’s first publicly available sovereign conversational AI infrastructure, and a milestone in the Group’s wider push to deepen these capabilities. By pairing open-source development with enterprise-grade deployment, Makimoto complements Toku’s existing customer experience platform and supports long-term innovation opportunities across the Group’s AI roadmap.
At launch, Kawa is available as a post-conversation transcription API, built for recorded calls, voicemail, and large-scale analysis of archived customer interactions. Developers submit an audio recording to the API and then retrieve the completed transcript, with speaker separation and segment-level timestamps. Each account includes a free monthly allowance of 1,000 minutes and an open-source interactive playground for testing before integration. A real-time transcription API, designed for live voice agents, captioning, and in-conversation analytics, will follow as a subsequent release.
The Kawa API runs on a composable orchestration layer, allowing developers to swap individual components in the pipeline, such as audio pre-processing or the speech-to-text model, without redesigning the surrounding system.
Hosted in Singapore, Kawa keeps customer audio and transcripts in-country, supporting organisations operating under Singapore’s Personal Data Protection Act (PDPA) and sector-specific frameworks, including guidelines issued by the Monetary Authority of Singapore (MAS). It is designed to meet growing demand among enterprises and regulated industries for AI infrastructure that gives them greater control over how data is processed, where workloads run, and how regulatory obligations are met. The architecture established for Singapore provides a foundation for future deployments in markets with similar data-residency and compliance requirements.
“We committed in April to putting Kawa in developers’ hands by 1 July, and today we have. Models are becoming a commodity, and the cycles are getting shorter, so today’s best choice is next year’s legacy. The advantage that endures is a composable orchestration layer that lets you replace and upgrade components without rebuilding the system each time, and that work, in the open, is what Makimoto is for,” said Thomas Laboulle, Founder and CEO at Toku.
Developer interest has continued to build since Makimoto was first announced in April. The managed API available from launch provides developers and enterprises with immediate access to a production-ready transcription pipeline, while Kawa’s open orchestration layer enables inspection, configuration, and future extensibility. Through subsequent releases in 2026, Toku intends to progressively open additional layers of the pipeline, moving towards greater community participation and modifiability.
The launch also marks early progress on Makimoto’s Singapore-based AI ecosystem Toku announced in April. The ten graduate roles, spanning engineering, developer relations and community, drew more than 400 applicants from universities and polytechnics across the region, with the inaugural cohort now being finalised for hire. The response reflects strong regional appetite to build production-grade open-source infrastructure rooted in Asia-Pacific.
Makimoto is operated through Makimoto Technology Pte Ltd, a wholly-owned subsidiary of the Toku Group, maintaining a clear distinction between community and enterprise editions of its software.
About The Author
