Skip to main content
Unpod is the agentic backend for conversational AI. Apps that talk, remember and act: voice, messaging, agents and workflows in one SDK, on a Context Engine that remembers every customer across every channel. Today, voice and the Chat API run in production, and voice agents remember past calls; WhatsApp, SMS, email and cross-channel memory are in build. Every team building an app that talks assembles these layers from six vendors. Unpod ships them as one system. Build the product, not the plumbing.

Talk

How the backend reaches the human.
  • Calling Engine - programmable realtime voice: numbers, calls, SIP, transfers, streams.
  • Channels - voice and chat today; WhatsApp, SMS and email in build.
  • Models & Router - STT, TTS and LLM for real-time turns, with automatic failover.

Remember

The Context Engine: what survives every call, message and model swap.

Act

What decides and performs the next step.
  • Agents - workers that communicate and act, spun up per task.
  • Agent Harness - tools, retries, guardrails, human handoff, evals and observability around every agent.
  • Flows & Journeys - durable workflows: steps, waits, outcome branching.
  • Actions - your APIs, called mid-conversation or fed every outcome.
  • Events - everything that happened, once.
  • Observability - call logs, recordings, transcripts and traces.

Status

Every Core Engine page carries a status line. Production means it carries production traffic today. In build means it is being built and is not available to use yet. Early access means it is open to early partners. In graphics, a Planned chip marks part of the design that no one can use yet. Statuses are as of October 2026; where an API is callable today, the page links the doc where it lives now.

The adoption path

First call → one agent → memory of past calls → one more channel → one flow → a reply wakes the journey → context across every channel. Start with the first call: Quickstart.