@dotey: Kimi Code and DeepSeek Harness should build GUI as soon as possible, support office tasks early, and aim to become general-purpose agents. Competing on TUI and Coding has no future — of course, coding is a foundational capability; if coding is poorly done, other tasks won't be done well either.
Summary
@dotey suggests Kimi Code and DeepSeek Harness develop GUI and support office tasks early, becoming general-purpose agents, and believes that only competing on TUI and Coding has no future.
View Cached Full Text
Cached at: 06/01/26, 03:07 AM
Kimi Code, DeepSeek Harness should develop GUI support as early as possible, and better support office tasks to become general-purpose agents.
Doubling down on TUI and Coding leads nowhere. Of course, coding is a fundamental capability — if coding isn’t done well, other tasks won’t be done well either.
踏雪寻仙 (@TaXue2025): Two more new contenders worth watching: Kimi Code and Grok Build. Both update rapidly and have great potential.
Similar Articles
Show HN: What should the GUI for AI agents look like?
MarbleOS is a GUI workspace for AI agents, providing visible files, tools, tasks, and outputs instead of burying everything in chat threads. A beta download and demos are available.
Show HN: Supapool – a Supabase per coding agent in ~400 ms
Supapool is a CLI tool that provides isolated, ephemeral Supabase instances for each coding agent or CI run, spinning up in ~400 ms to enable parallel, production-like testing without interference.
@Saboo_Shubham_: You can now agent development lifecycle from inside your coding agent. Any coding Agent + Agents CLI + Agent Developmen…
A new workflow allows developers to perform the full agent development lifecycle (build, test, deploy, evaluate, govern) directly from within their coding agent, using Agents CLI and Agent Development Kit.
OmegaUse-OfficeVal: Benchmarking LLM Agents on Long-Horizon Office-Suite Tasks with Economic Grounding
OmegaUse-OfficeVal is a benchmark for evaluating LLM agents on long-horizon office-suite tasks with economic grounding, comparing human costs and LLM inference costs. It includes 100 tasks requiring ~2.3 hours of human labor each, and finds that frontier LLMs are cheaper and faster but still below human quality.
We rewrote our agent to run entirely in a Durable Object with Pi, Agents SDK, and Code Mode (10 minute read)
camelAI rewrote their coding agent to run entirely in a Cloudflare Durable Object, using SQLite and R2 for filesystem and replacing bash with JavaScript. The migration from VMs reduced costs and latency, and the codebase is now open source.