OpenClaw Ships Mobile Companion Nodes for Agent Gateways
The new OpenClaw iOS and Android applications function as companion nodes that connect mobile hardware capabilities to self-hosted AI agent gateways.
On June 29, 2026, the open-source platform OpenClaw released native mobile applications for iOS and Android, followed immediately by the v2026.6.11 patch to stabilize gateway connectivity. The release transitions the framework from a desktop-bound environment into a distributed agentic network capable of utilizing mobile hardware.
Gateway Architecture and Device Integration
Unlike standard standalone chatbots, the OpenClaw mobile applications do not execute AI models locally. They operate as companion nodes within a self-hosted agent network. If you build advanced agents with OpenClaw, your mobile device now pairs with a private OpenClaw Gateway running on macOS, Linux, or Windows/WSL2 via a setup code or QR code.
The application connects to the gateway using a WebSocket on port 18789. This architecture grants the remote agent authorized access to on-device capabilities. The mobile node exposes the camera, location data, photos, contacts, calendar, and reminders directly to the agent’s context window. The integration includes a Talk mode for real-time and background voice sessions alongside push-to-talk functionality.
Version v2026.6.11 Patch Details
Initial deployments revealed UI overlapping bugs and connection drops. The OpenClaw Foundation shipped release v2026.6.11 less than 24 hours after launch to address these stability issues. The update implements TLS hardening and centralized provider transport to secure the connection between the mobile node and the host.
| Component | v2026.6.11 Improvement |
|---|---|
| Channel Reliability | Fixed misplaced and dropped replies across Telegram, WhatsApp, iMessage, and Google Chat. |
| Execution Pipeline | Resolved stalled requests and stuck sends during complex task execution. |
| Node Visibility | Added Android detail panels for real-time status monitoring. |
Enterprise Adoption and Heritage
The project originally operated under the name Warelay before its rapid expansion under founder Peter Steinberger. Microsoft validated the underlying framework on June 2, 2026, when it announced Microsoft Scout, a new autopilot system built on OpenClaw’s open-source technology.
The decoupling of the interface from the execution environment resolves the friction of getting autonomous agents through strict mobile app store reviews. Developers can now run multi-tenant agents on a private server and implement multi-agent coordination patterns while exposing the interface securely to a phone.
If you are deploying a mobile node, confirm your gateway host firewall explicitly allows inbound TCP connections on port 18789 and configure a reverse proxy to handle the TLS termination before routing traffic to the WebSocket.
Get Insanely Good at AI
The book for developers who want to understand how AI actually works. LLMs, prompt engineering, RAG, AI agents, and production systems.
Keep Reading
How to Run Gemma 4 On-Device with LiteRT-LM
Learn how to configure LiteRT-LM to deploy the Gemma 4 model family locally across mobile, desktop, and edge environments with constrained JSON decoding.
Multilingual PP-OCRv6 Beats GPT-5.5 on Industrial Text
PaddlePaddle's PP-OCRv6 system delivers 50-language text recognition in a 34.5M parameter footprint that outperforms massive vision models.
Local LLMs Hit 88% Accuracy Triaging OpenClaw Pull Requests
Learn how to build a local AI triage system for GitHub repositories using Hugging Face Text Generation Inference, smolagents, and Command R.
$99 Google Home Speaker Retires Assistant for Gemini
Google launched the $99.99 Google Home Speaker, featuring a dedicated NPU to shift the ecosystem from legacy Assistant to the generative Gemini for Home.
MIT-Licensed GLM-5.2 MoE Reaches 74.4% on FrontierSWE
Zhipu AI has released GLM-5.2, a 744-billion parameter MoE model engineered for long-horizon agentic tasks with a stable one-million-token context window.