March 26, 2026
Wrapper wars, emoji outrage
Show HN: Relay – The open-source Claude Cowork for OpenClaw
Open‑source “Cowork” clone drops—HN erupts over wrappers, emojis, and “AI slop”
TLDR: Relay promises a local, open-source command center for AI assistants with audit trails and model choice, aiming to replace cloud‑hosted workflows. The HN crowd fires back over “wrapper-on-wrapper” bloat, questions claims about Anthropic’s setup, and grumbles that “AI slop” keeps hitting the front page—drama included.
Relay just hit Hacker News promising a local control center for AI helpers—think Claude Cowork’s workflow, but running on your own machines, with your own models, and approval/audit logs baked in. The pitch: keep your data at home, ditch lock‑in, schedule tasks, and route to anything from GPT‑4 to Llama.
But the comments? Spicy. One top take calls Relay “a wrapper on top of a wrapper… on top of the API,” blasting the stack as overweight when you could just call the service directly. Others didn’t argue architecture—they went for vibes. A long‑time user slammed the README’s emoji/ASCII flair as “disrespectful,” reading it as marketing gloss over substance. Cue the meme flood: “wrapperception,” “audit logs but make it ✨,” and “ASCII art is the new compliance.”
Then comes the fact‑check scuffle: a commenter says claims about Anthropic’s Cowork running files on Anthropic servers are “misleading,” challenging Relay’s comparisons. Meanwhile, another thread explodes with “Why does AI slop rank so high?”—only to be met with a sharp “Says a 2‑day old account? Nice” clapback. Translation: credibility wars and karma conspiracies.
Bottom line: Relay’s promise of control, choice, and compliance is clear—but the community is split between excitement for governance features and suspicion that it’s just wrappers on wrappers with a shiny README.
Key Points
- •Relay is an open-source Electron desktop app that serves as a control plane for OpenClaw, providing a local alternative to Claude Cowork.
- •It addresses data sovereignty, model lock-in, and compliance gaps by running on user infrastructure, supporting multiple LLMs, and offering exportable audit trails with approval gates.
- •Relay manages task dispatch, chat, governance approvals, cost tracking, scheduling, agent memory, connectors (Slack, Notion, GitHub, Jira), and project folders.
- •OpenClaw functions as the execution plane with autonomous agents, sub-agent orchestration, persistence (memory, cron, file I/O), and integrations (connectors, browser/UI, shell/scripts).
- •The model router enables routing to various LLM backends, including Claude, GPT-4, Llama, Gemini, Mixtral, and custom endpoints, with an example use case of a scheduled daily briefing.