SautiMesh v0.1.0 — Voice-First Offline Multilingual AI Agent Runtime

SautiMesh v0.1.0 — Voice-First Offline Multilingual AI Agent Runtime

We are excited to announce the initial release of SautiMesh (v0.1.0), an offline-first, voice-first agentic runtime designed specifically for Africa and edge deployments worldwide.

Key Capabilities

• 8 African & Global Languages Supported: First-class support for Swahili, Luganda, Yoruba, Hausa, French, English, Nigerian Pidgin, and Runyankole, complete with seamless multilingual and code-switching intent recognition.
• Voice-First Audio Pipeline: 16kHz PCM Voice Activity Detection (VAD) with energy & zero-crossing rate thresholding, circular buffering, and low-latency audio capture.
• Store-and-Forward Durability: Embedded SQLite WAL persistent message queue with priority dispatch, exponential backoff with jitter, and dead-letter queueing for intermittent connectivity.
• Bandwidth & SMS Optimization: GSM-7 7-bit septet packing and UCS-2 16-bit encoding with multi-part concatenation headers for edge transport.
• Stdio MCP Integration: 5 stdio Model Context Protocol (MCP) server tools (sautimesh_voice_in, sautimesh_transcribe, sautimesh_dispatch, sautimesh_sms_out, sautimesh_status) ready for Claude Desktop, Cursor, and agentic workflows.
• 100/100 Test & Audit Verification:
  • Full pytest test suite passing with 100% coverage across core modules.
  • BillShock audit passed with zero unaccounted cloud egress.
  • Potemkin audit passed confirming zero mocked or hollow implementations.
  • GrowthPilot verification passed.

Documentation & Live Studio

• Website & Interactive Studio: https://codesphere-u.github.io/sautimesh/
• Architecture Docs: https://codesphere-u.github.io/sautimesh/docs.html
• Interactive Studio: https://codesphere-u.github.io/sautimesh/studio.html