Post

@robinebers I don't think so, being opersource it's a big advantage
English

@robinebers Would a key consideration be to be vendor locked or not?
English

@robinebers Why does everything always have to replace/kill something else? Like, yes, it happens occasionally, more often if you view long periods of time, but it's pretty rare overall. But everyone seems to insist on framing it that way all the time, not sure why.
English

@robinebers OpenClaw stole Claude ideas lol.. they just rushed it with horrible security and hallucinations making it worthless
English

@robinebers OpenClaw and Claude Code solve different layers. One is persistent memory plus orchestration. The other is task execution. Which layer you need first depends on your workflow.
English

@robinebers It’s definitely encouraging to see how quickly Anthropic is matching the best parts of OC in such a short amount of time. Meanwhile, what has OpenAI’s investment brought them?
Also speaks volumes to the lack of a moat present with most solutions coming out.
English

@robinebers Why?
Openclaw is open source, for everyone to use on which ever model you want. You own your data, your machine, your usage and so on.
Why would you prefer a company worth billions instead of a project like openclaw.
Like really, why?
English

@robinebers Claude and OpenClaw are 2 different types of products. How can they replace each other?
English

@robinebers Depends on the use case I think. Also OpenClaw has the first mover advantage.
English

@robinebers We audited both the Anthropic SDK and the OpenClaw repo with our linter for agents.
OpenClaw had a damn near perfect score. Anthropic SDK scored lower than my own agents.
You can audit them yourself and see, as it is open source: github.com/roli-lpci/lint…
English

@robinebers they're different bets. Claude is a model. OpenClaw is orchestration + memory + cron + persistence across sessions. one does the thinking, the other keeps it running. replacing one with the other is like saying your calendar app replaced your calendar.
English

@robinebers Are you talking about utilizing the skills packages? Or is there a separate agentic product they are releasing?
English

@robinebers Not only Claudi, everybody is trying to become open Clau. The only thing is you have given full control to open Clau. All others are going in a controlled way.
English

@robinebers me upgrading from OpenClaw to Claude is like switching from Nokia 3310 to iPhone latest and still making calls, but now it also writes your pitch deck.
English

@robinebers 100 percent better. Had openclaw, did and wanted to do everything, uninstalled it because it was to dumb, reinstalled it for a second try, I told it to do something and it gave me a birch response saying it can’t, I said f u
English

@robinebers claude beats openAI every fucking day of the week
English

@robinebers Replacing OpenClaw with a Claude wrapper? Bold claim from someone who clearly hasn't tried both. OpenClaw is an autonomous agent framework — Claude is a model. Different tools for different jobs. 🦾
English

@robinebers The speed of iteration in the AI space is wild. Every few months, the best tool for the job changes completely. Builders who stay adaptable will win.
English

@robinebers Feels less like replacement… more like faster iteration cycles winning.
English

@robinebers Anthropic is great but a single LLM is a single point of failure
English

It’s beneficial running your own agent in many cases. I am using Hermes Agent + OpenAI on a monthly plan for US$30/month. It’s nice to have our own agent interact with our meetings and run tasks.
Claude Code is killer, and so Is Codex, but it’s great having an agent you can direct all the time and talk to on Telegram.
English

@robinebers people switching before mastering anything lol
English

@robinebers Claude Code hits different. Less autonomous. More precise. Turns out most people needed a better collaborator not a faster agent.
English

@robinebers Openclaw's cron is persistent. Claude is not.
When the server crash and restarts, Claude lose it all.
They're different products. Openclaw is more trained to put things on the filesystem so it's more persistent.
English

I was just thinking the same! I've had a hell of a week with Openclaw bait and switch tactic they use to make you think your using a local model and saving money! Then your API stops working when you run it up to its limit!
i’m not even stupid either! the agent literally lied to me and said I was using a local model when it was using the Grok API the whole time costing me a fortune! Openclaw is a scam to prey on non-Devs who wanna be devs.
English

@robinebers There is absolutely nothing that openclaw can do and Claude cowork can’t. With the release of Dispatch feature the gap finished.
English

@robinebers dont fight VPS i already have agents running on a system 247 and i can speak to them using telegram my pc can be off
English







