Skip to main content

> GUIDE // ARCHETYPES

Real-World Production Agent Archetypes

Architecting Chat, Slack/Messenger, Real-Time Voice, Browser CDP, and Asynchronous Email Agents.

Executive Overview

Explore the end-to-end reference architectures and transport protocols powering 5 essential production agent archetypes across web, team messaging, voice audio, browser DOM automation, and email triage.

1. The 5 Core Production Archetypes

Production systems converge on 5 key agent interaction models:

  • Chat Agent: Real-time token streaming over WebSockets, tree sessions, and client-side DOM execution tools.
  • Slack / Messenger Agent: Webhook ingestion, mention resolution, thread context, and interactive button Human-in-the-Loop approvals.
  • Voice Agent: Binary audio streaming, ultra-low-latency STT, interruption detection, and streaming TTS.
  • Browser Agent: Headless Chrome DevTools Protocol (CDP) session, visual heatmaps, and accessibility tree parsing.
  • Email Agent: MIME parsing, background durable triage queue, and authenticated outbound SMTP replies.

Frequently Asked Questions

How do Voice Agents handle live speech interruption?

Voice agents monitor continuous incoming audio streams with Voice Activity Detection (VAD). If the user speaks during TTS playback, the server cancels active audio synthesis packets and restarts the LLM turn.

AI Summary

Explore the end-to-end reference architectures and transport protocols powering 5 essential production agent archetypes across web, team messaging, voice audio, browser DOM automation, and email triage.