Phonevoice, voice agents in production
Call handling, audio streaming, transcription, routing and business integrations. We meet the demands of production telephony every day, on our own service.
We run it too. Four products of our own, in production, with their customers, their incidents and their releases. That is what keeps us honest about what a project really costs once it has shipped.
AI voice agents for enterprises: qualification, routing and call handling, real-time audio, CRM integrations and human supervision.
Voice-first AI assistant. Real-time voice, email, calls, web agent, multi-LLM. Cloud commercial + open-source self-hosted edition.
Task orchestrator for AI coding agents. Workspace, claims, runs, and a metered AI proxy. CLI + SDK open source, cloud is paid.
Universal linter with AI-powered code review. One CLI for ESLint, RuboCop, Ruff, Biome, Stylelint and more. Open source.
These are not demos. They are products we operate: authentication, billing, monitoring, support, releases and roadmap trade-offs.
Call handling, audio streaming, transcription, routing and business integrations. We meet the demands of production telephony every day, on our own service.
Real-time speech, calls, web agents, self-hosted mode and device firmware: the full chain of a voice assistant, from the microphone to the model.
Workspaces, runs, CLI and SDK, a paid tier: the machinery of a multi-tenant service — the same machinery we put in place for clients.
An open npm package with AI code review, several linters and careful CI ergonomics. Shipping a tool other people rely on sets its own bar.
Each product follows the same pragmatic playbook: short framing, readable architecture, progressive delivery, observability and continuous maintenance.
We bring the same level of rigor we use on our own products to client work — voice agents, hardware, AI-heavy SaaS.