Toku has launched Makimoto Flows, a visual builder that lets developers assemble, test and export speech AI pipelines without manually wiring every processing stage in code. The tool extends the Singapore-headquartered company's Makimoto initiative beyond transcription, combining speech-to-text, call summarization and conversation tagging in a configurable workflow. Available at makimoto.ai, Flows gives developers a way to inspect how recorded calls move through each stage while keeping processing within Singapore, according to Toku.

Toku expands Makimoto from transcription to speech AI orchestration

The launch marks a shift in Makimoto's development strategy. Rather than offering transcription as an isolated service, Toku is building a modular stack in which developers can connect speech processing and analysis components into workflows tailored to specific customer experience applications.

Flows includes three main blocks. Transcribe converts recorded audio into text using Kawa, Makimoto's speech-to-text service. Summarise generates a topic and short summary from the transcript, while Tags classifies calls across seven categories, including customer sentiment. Developers can inspect individual stages, modify the pipeline and rerun recordings to compare outputs.

A Chained Analysis template combines the blocks into a ready-made workflow. Each block maps to a Makimoto API endpoint, and developers can export a completed flow as executable Python code. This creates a bridge between visual experimentation and software development: teams can prototype a call-analysis process in the browser, then incorporate the same capabilities into their own applications.

The approach could support practical contact-center tasks such as generating call recaps, sorting conversations for follow-up and identifying interactions that require additional attention. However, the announcement does not establish independent accuracy benchmarks for transcription, summarization or sentiment classification.

Open architecture meets data residency requirements

Makimoto builds on Toku's April 2026 launch of its conversational AI initiative. Kawa became available through a Singapore-hosted API on July 1, followed by summarization and tagging endpoints in September. Flows adds a visual orchestration layer to those existing capabilities.

The Singapore deployment is an important part of the proposition. Toku says every Flows block runs in Singapore, keeping audio, transcripts and analysis within the country. That may appeal to enterprises handling customer conversations that contain personal or commercially sensitive information.

Data residency, however, is not equivalent to complete regulatory compliance. Businesses still need to evaluate access controls, retention policies, consent requirements, subprocessors and how data moves through downstream systems.

The initiative also emphasizes open-source development and composability. Instead of tying the entire workflow to a single model or fixed sequence, the architecture is intended to let developers change individual components as requirements evolve. Toku's stated roadmap includes audio preparation blocks, beginning with noise filtering, before transcription takes place.

Competing in the conversation intelligence market

Flows enters a market that includes contact-center analytics, speech-processing APIs, conversational AI frameworks and customer experience platforms. Established cloud providers such as Amazon Web Services and Google Cloud offer speech and AI infrastructure, while specialized contact-center vendors provide analytics and workflow automation integrated with their broader platforms.

Makimoto's differentiation is its combination of visual pipeline composition, API-level access, Python export and Singapore-based processing. Developers can experiment with a workflow without first writing the full integration, yet retain a route to custom code.

That positioning is distinct from a complete contact-center platform. Flows is a builder for speech AI pipelines, not evidence of a replacement for enterprise CRM, workforce management or omnichannel customer engagement systems. Its value will depend on how easily it connects to those environments and whether its outputs are reliable enough for production use.

Toku is offering developers 1,000 free minutes per account, lowering the initial barrier to testing the platform. The next challenge is to turn experimentation into sustained adoption among engineering teams and enterprise buyers.

For CX leaders, the relevant question is whether configurable speech AI can make customer conversations more actionable without adding operational complexity. Flows gives developers a new way to build and test those workflows; production reliability, integration effort and measurable business impact will determine whether it becomes a meaningful part of enterprise CX infrastructure.

Market Landscape

Conversation analytics is becoming a core part of customer experience infrastructure as businesses seek to extract structured insights from recorded customer interactions. Gartner's Market Guide for Conversation Analytics Platforms, published in November 2025, identifies this as a distinct technology category for customer service and support leaders evaluating ways to analyze customer conversations.

Makimoto Flows sits at the intersection of three markets:

• Speech AI infrastructure: APIs and models that convert audio into text and support downstream processing.

• Conversation intelligence: Tools that summarize calls, classify interactions and surface customer sentiment.

• Developer orchestration: Visual builders and composable frameworks that connect AI components into custom workflows.

Its competitive advantage will depend on more than the visual interface. Developers will need reliable multilingual transcription, consistent outputs, flexible integrations and transparent data-handling practices. Singapore-based processing may be particularly relevant to organizations with regional data-residency requirements, although buyers should validate their specific compliance obligations.

Strategic Outlook

Toku's roadmap points toward a more configurable speech AI stack, with audio preparation and noise filtering planned as the next additions. If delivered, these components could give developers greater control over the quality of recordings before transcription and analysis.

The broader industry direction favors modular AI infrastructure that lets businesses change models and processing stages without rebuilding entire applications. For enterprise CX teams, this can reduce dependence on rigid workflows, but it also introduces responsibility for testing, monitoring and maintaining each component.

The key milestones for Makimoto will be production-grade reliability, integration with existing contact-center systems, transparent performance benchmarks and adoption beyond developer trials.

Top Insights

• Visual pipeline building: Makimoto Flows lets developers connect transcription, summarization and tagging blocks, making speech AI workflows easier to prototype and inspect.

• Singapore-based processing: Toku says audio, transcripts and analysis remain in Singapore, addressing a key infrastructure consideration for organizations with regional data-residency requirements.

• Code export supports deployment: Python export allows developers to move tested visual workflows into custom applications, bridging experimentation and production integration.

• Open architecture offers flexibility: Makimoto's composable design aims to let teams change processing components independently as speech AI models and business requirements evolve.

• Free access lowers trial barriers: The 1,000-minute allowance gives developers a practical way to test recorded-call analysis before evaluating enterprise deployment requirements and costs.

Get in touch with our MarTech Experts.