# Show HN: Designing the GUI for AI Agents > Source: > Published: 2026-08-19 13:36:56+00:00 A Show HN thread titled *"What should the GUI for AI agents look like?"* sparked debate among developers, designers, and ops engineers. The post displayed mock‑ups and asked for feedback on layout, interaction, and the trade‑off between low‑code flexibility and technical depth. Contributors favored node‑based canvases, timeline views, and form‑driven panels. They also warned about scalability, state visibility, and debugging support. The thread gathered dozens of comments. Suggestions included collapsible sub‑graphs for complex hierarchies, real‑time telemetry overlays, and drag‑and‑drop parameter binding. No single design won, but the consensus points to a hybrid UI that mixes visual flowcharts with contextual property editors. The goal is to make AI‑agent orchestration intuitive and powerful for production teams. **Q: How can I integrate a new AI‑agent UI into an existing n8n instance without breaking current flows?** A: Encapsulate the UI as a separate plugin or custom node library. Use feature flags to enable it for specific users, and keep the underlying workflow JSON format unchanged so existing flows keep running. **Q: What’s the best way to surface runtime metrics (like token count) in the GUI without overwhelming the canvas?** A: Add a lightweight overlay panel that appears on node hover or via a toggle button. Show key metrics in a concise badge (e.g., "🟢 120 tokens") and let users expand the panel for deeper logs. **Q: Should I design the GUI for AI agents as a single monolithic view or break it into multiple tabs?** A: Use a tabbed approach for distinct concerns—"Design", "Parameters", "Telemetry", and "Permissions". This keeps the main canvas clean while still giving power users quick access to advanced settings. *Originally published on Automations Cookbook.*