proomt

Search

Search posts, papers, and topics

All posts

GitHub OldBurke Holland5 min readintermediate

When chat is the wrong UI

Summary

This article argues that chat is often the wrong UI for interacting with LLMs, especially for task-oriented work. It introduces GitHub Copilot's "canvases" as customizable, full-stack applications that run within the Copilot app, enabling more efficient, specialized, and automated interactions with AI agents.

  • Chat is a universal but often inefficient UI for LLMs, particularly for known, repetitive tasks.
  • GitHub Copilot's "canvases" are full-stack applications that run inside the Copilot app.
  • Canvases support bi-directional communication with agents, third-party API calls, and local code execution.
  • They allow agents to build specialized tools for tasks, reducing token waste compared to constant chat interactions.

Developers using LLMs for specific tasks should consider custom UIs like GitHub Copilot's canvases to improve efficiency, reduce token usage, and automate complex workflows beyond simple chat interactions.

7/10

Related reading

  1. Verifiable Social Reasoning for LLM Assistants

    The paper introduces Fuse, a multi‑agent simulation that gives LLM assistants a verifiable ground‑truth task for social reasoning by hiding a target agent’s motive and letting a user‑mediated conversation infer it. Experiments on 12 LLMs show user mediation makes reasoning harder, models are biased by user framing, need more detail than humans, and longer chats don’t always help.

    Hugging Face Daily Papersarxiv.org1 minpaper