LiveKit Agents vs Spinach
Side-by-side trajectory, velocity, and editorial themes.
LiveKit Agents added Answering Machine Detection — voice agents are becoming a serious telephony runtime.
LiveKit Agents is releasing roughly twice a week along the 1.5.x line, accumulating telephony-grade primitives around its voice loop. The headline is Answering Machine Detection in 1.5.9 — an LLM-classified detector for what kind of endpoint an outbound call hit. Surrounding work is split between reliability (barge-in cooldown, interruption guards, preemptive-generation tuning, observability retries) and provider breadth (Perplexity Responses, Soniox, Speechmatics, Cerebras, xAI, Rime WebSocket TTS). The mcp_servers parameter was also deprecated on Agent and AgentSession.
The product is converging on a real contact-center runtime, not just a realtime meeting agent. AMD, warm transfer, DTMF handling, recording retries, and avatar join/playback metrics are the feature surface phone deployments demand. The provider plugin universe keeps widening; LiveKit positions itself as the neutral broker between voice models and the actual network. Internal cleanups (mcp_servers deprecation, instruction parts, AvatarSession base class) suggest a tidying pass before a 1.6 cut.
Expect more telephony primitives — supervisor barge-in, richer DTMF flows, call-recording controls — and a unified MCP configuration surface across Agent and Session as the mcp_servers deprecation lands fully.
Filling out the meeting-transcript-to-AI-agent integration matrix, one connector at a time.
Spinach is publishing a tightly coordinated content matrix: how to pipe Zoom, Google Meet, and Microsoft Teams transcripts into every major AI workspace and dev tool. Two date clusters dominate — five posts on April 24 and five more on May 1 — each running the same template across a different combination of source meeting platform and destination agent (Claude Code, Claude Cowork, Codex, Glean, Notion AI, HubSpot, Linear).
Spinach is repositioning from "AI meeting assistant" to "transcript pipeline for the rest of your AI stack," with its MCP server as the underlying connective tissue. The choice of destinations is telling — heavy emphasis on engineering tooling (Claude Code, Codex, Linear) suggests the GTM is moving toward technical buyers rather than the original ops/PM audience.
Expect more matrix entries — Cursor, Devin, JetBrains AI, ChatGPT desktop, Salesforce — published in fast batches. A consolidated "integrations directory" or marketplace page is the natural next visible artifact.
See more alternatives to LiveKit Agents →
See more alternatives to Spinach →