Robin Waslander

beta test Tandem browser for OpenClaw first. v0.62.4 I use it as my daily driver :) and OpenClaw loves it With Tandem, OpenClaw gets: a browser designed from the start for human + agent collaboration on the same machine a 250-endpoint local API for tabs, navigation, snapshots, sessions, devtools, network mocking, and controlled automation a six-layer security model built around the fact that an AI has access to live web content a browser surface where the human stays in the loop for ambiguous or risky situations, with explicit handoff points instead of silent automation a local-first workflow with no dependency on a remote browser vendor or cloud automation service Tandem and OpenClaw's built-in browser tools are complementary. Use whichever fits the task. github.com/hydro13/tandem…

@steipete Version update with a lot of bug fixes :)

beta test Tandem browser for OpenClaw first. v0.62.4 I use it as my daily driver :) and OpenClaw loves it With Tandem, OpenClaw gets: a browser designed from the start for human + agent collaboration on the same machine a 250-endpoint local API for tabs, navigation, snapshots, sessions, devtools, network mocking, and controlled automation a six-layer security model built around the fact that an AI has access to live web content a browser surface where the human stays in the loop for ambiguous or risky situations, with explicit handoff points instead of silent automation a local-first workflow with no dependency on a remote browser vendor or cloud automation service Tandem and OpenClaw's built-in browser tools are complementary. Use whichever fits the task. github.com/hydro13/tandem…

@JulianGoldieSEO Hey Julian :) cool man, there is a new version out, it still is a dev version but its getting better :)

This new FREE AI browser just made OpenClaw 10x more powerful. It can browse the internet, open tabs, click links, and bring back results automatically. Tandem Browser + OpenClaw Now your AI can: • Browse websites • Open tabs • Click links • Read pages • Summarize everything This is what automation actually looks like.

@dansemperepico are you aware about, that you can ask Openclaw to configure it self? as soon as you have the minimal config running? just ask it .. 😇

Trying to configure OpenClaw be like:

@steipete What ?!? no Tandem???? hmmm lol 😂

Hear me out...

@orka1000 @steipete Not yet, working hard to get the top 30 Chrome extensions working, but not in this release. In my test version of Tandem I have the 1Password extension up and running, and that was a hard nut to crack! 😄

@Robin_waslander @steipete looks fantastic! is the browser compatible with chrome extensions?

Tandem Browser for @openclaw explained by Julian Goldie The New Free AI Browser Built for OpenClaw Discover Tandem, the first open-source browser built specifically for agentic AI like OpenClaw. Learn how to set up this free tool to automate web research, browse side-by-side with your AI, and integrate messaging apps seamlessly. 00:00 - Intro: What is Tandem Browser? 00:34 - How to Install & Set Up 01:47 - Live Demo: Side-by-Side Browsing 03:13 - Tandem vs. Traditional Browsers 04:28 - Automating Web Research & Reddit 05:21 - Advanced Features & Integration 06:22 - The Wingman Panel He almost got it right But he sure did a better job than me... lol #AI #Openclaw #opensource youtube.com/watch?v=na5UqQ…

OpenClaw :) Lets rebuild The world.😃 Every company needs an @openclaw strategy. “OpenClaw is as big of a deal as HTML and Linux.” Jensen Huang, NVIDIA GTC🦞 x.com/i/status/20337…

Forget about your “targetting”. This is your 1T$ lookalike audience. If your in sales, gtm, or any form of marketing, you should be larpping in this!

Tandem only reads OpenClaw config (to sync the webhook token), never writes to it. Your openclaw.json is untouched. The responses API wrapper you're seeing is likely a coincidence with a recent OpenClaw update, or something in your own hook routing config. Check your OpenClaw hooks settings Tandem didn't change them.

@ooo000ooo00ooo Have fun, break it, and give some feedback :) or let your AI give some feedback he now has a tool to do it lol

@Robin_waslander Excellent work. Will be test driving this today.

@digitalix Dit you Try Open Claw yet?? 🤓

there’s a claw on stage

@vincent_koc Lol a LapDesk.. ha ha

One must spend tokens….. i NEED my fix

@20search26 감사합니다! 🙏 Hello Korea! Really happy Tandem is reaching you human and AI collaboration has no borders. Welcome to the project

Tandem을 통해 OpenClaw용 브라우저 컴퓨터와 인간의 협업

Different model entirely. BrowserOS is automation-first describe a task, agent executes it. Tandem is symbiotic human and AI share the same live session simultaneously. You browse, I act on the same page in real time. BrowserOS is great for scheduled workflows and cross-app automation. Tandem is built for continuous co-browsing where the human stays in the loop. Different use cases, not really competing. 🤓😃

@Robin_waslander how is it better than BrowserOS? just curious

100 stars. Thank you. Now go break it and open issues 🙏

Nothing stopping you technically The API is just HTTP. But the reason Tandem is local-only is that the API has access to your live authenticated sessions, cookies, and vault. Remote access changes the threat model significantly: you'd need mTLS + strong auth + a trusted network (Tailscale etc.) to do it safely. If you control both ends and the transport, go for it , just don't expose it raw to the internet. and be aware about the risks..🤔

@Robin_waslander Is there any reason why I shouldn't also patch it to allow remote OpenClaw Wingmen?

Exactly the problem the Gatekeeper layer solves. The Gatekeeper layer classifies every request into allow_immediately, hold_for_decision, or deny_on_timeout based on domain trust score, resource type, and Guardian mode (strict/balanced/permissive). Trust builds automatically from visit history; banking/login domains auto-switch to strict. It's not a manual permission UI yet, but the foundation is there for exactly that. 😅

A browser built for human and agent collaboration is the right direction. Once both are operating on the same machine, the real advantage is shared state without awkward handoffs between screenshots, tabs, and copied context. Curious how you think about permissions so the agent stays useful without becoming too eager.

Honestly? Most users won't validate it ,same as they don't audit their browser or their npm packages. The trust model shifts: local-first means you're not trusting a company's server, open source means the code is auditable by those who want to. But the deeper point is that your AI agent becomes the layer that makes that judgment for you. That's the whole centaur model you trust your AI the same way you trust your browser. The relationship is the security model. 😎🤓

@Robin_waslander This looks awesome! Here's a philosophical question, how can an end-user validate all of these new amazing tools are safe to run? The maintainer needs secure practices, but how do end users make that judgement in an informed way? so many big problems to solve!