Anthropic shipped Claude Dispatch on March 17, turning the Claude mobile app into a remote control for your Mac’s Cowork agent. Send a message from your phone, and Claude gets to work on your desktop — pulling files, querying databases, building reports. When it’s done, you get a notification. The idea is simple: your AI keeps working even when you walk away from your desk.
Within 24 hours, Dispatch hit #1 on Product Hunt with 470+ upvotes. TechCrunch, MacStories, and blockchain.news all covered it. On Twitter/X, the phrase “phone-to-desktop AI” started trending among developer circles. But the early hands-on reviews tell a more nuanced story than the hype suggests.
How Dispatch Actually Works
The setup is straightforward. After updating the Claude Mac app, a new “Dispatch” option appears inside Cowork. Tap it, scan a QR code with your iPhone (or Android device), and a persistent conversation thread syncs between your phone and desktop. From that point on, anything you type into the mobile app gets routed to the Cowork agent running on your Mac.
The connection uses end-to-end encryption, so the instructions traveling between your phone and computer stay private — Anthropic says they cannot read the content of dispatched tasks. Code execution happens inside a sandbox on your local machine. Files never leave your computer. And every action Claude takes requires explicit user confirmation before it proceeds.
This is Anthropic’s core pitch: the same remote-control convenience that open-source alternatives offer, but with enterprise-grade security guardrails baked in.
What It Can (and Can’t) Do
MacStories published one of the first detailed hands-on reviews, and the results were mixed. Here’s what worked:
- File retrieval: Dispatch successfully located screenshots, documents, and local files on the Mac
- Database queries: It accessed Notion databases through Connectors and summarized the contents
- Note summarization: Pulling together information from scattered notes worked reliably
And here’s what didn’t:
- Opening applications: Dispatch couldn’t reliably launch apps on the desktop
- Sending messages: Attempts to send Slack or email messages through Dispatch failed
- Terminal access: No ability to run shell commands or interact with terminal sessions
- Third-party auth: Services requiring OAuth or login credentials were a dead end
The MacStories reviewer described reliability as “about 50/50” and called the speed “slow.” The honest assessment: Dispatch is “not good enough to rely on when you’re away from your desk” in its current state — though the reviewer acknowledged it’s a research preview, not a finished product.
There are also some practical constraints worth noting. Your Mac must stay awake with the Claude app open. There’s no background service that keeps running if you close the lid. And Dispatch currently supports only single-threaded execution — you can’t fire off multiple tasks in parallel.
The OpenClaw Elephant in the Room
It’s impossible to talk about Dispatch without mentioning OpenClaw, the open-source desktop AI agent that has dominated this space for months. Peter Steinberger’s project racked up over 320,000 GitHub stars in roughly two months — making it one of the fastest-growing open-source projects in history. OpenClaw works with any LLM, connects via Telegram, WhatsApp, or Discord, runs 24/7 as a background service, and has a marketplace (ClawHub) with over 3,000 community-built extensions.
Latent Space’s AI newsletter explicitly framed Dispatch as “Anthropic’s Answer to OpenClaw.” And the comparison highlights some stark trade-offs:
| Claude Dispatch | OpenClaw | |
|---|---|---|
| Price | $100–200/mo (Max), $20/mo (Pro, coming soon) | Free (open-source) |
| LLM support | Claude only | Any LLM |
| Messaging | Claude mobile app only | Telegram, WhatsApp, Discord |
| Background running | No (Mac must be awake, app open) | Yes (runs as a service) |
| Extensions | Connectors (limited) | 3,000+ on ClawHub |
| Security | Sandboxed, E2E encrypted, user confirmation | Community-audited, known vulnerabilities |
| Task scheduling | Not supported | Cron jobs built-in |
But security is where Dispatch has a genuine edge. OpenClaw has faced serious scrutiny: Palo Alto Networks called it “the potential biggest insider threat of 2026,” AI researcher Gary Marcus described it as “a disaster waiting to happen,” and China’s CNCERT issued warnings about prompt injection risks. OpenClaw has no bug bounty program and no dedicated security team.
Anthropic is essentially betting that a subset of users — particularly those in enterprise, finance, and healthcare — will pay $100–200/month for the peace of mind that comes with sandboxed execution, encrypted communication, and a company standing behind the security model.
Who Should Care About This
Dispatch makes the most sense for people already deep in the Anthropic ecosystem. If you’re a Claude Max subscriber ($100/month for 5x usage or $200/month for 20x) using Cowork daily, Dispatch is a natural extension — your existing workflows, Connectors, and conversation history all carry over.
For knowledge workers who spend their days in documents, spreadsheets, and databases, the “text Claude a task from your commute” workflow has real appeal. Think: “Summarize the Q1 sales data and put it in a new doc” or “Find all invoices from February and flag anything over $10K.” These are the kinds of tasks where Dispatch’s current capabilities — file access, database queries, document generation — actually shine.
But if you need an always-on agent that handles complex automation chains, works with multiple AI models, or integrates with a dozen messaging platforms, OpenClaw (with its security risks acknowledged) or similar open-source tools are still more capable today.
Pro subscribers at $20/month will get access “within a few days” according to Anthropic, and a broader rollout to free-tier users is planned for later in 2026.
The Bigger Picture
Dispatch signals something important about where Anthropic is heading. The company has been methodically building out an agentic AI stack — Claude Code for developers, Cowork for general productivity, and now Dispatch for cross-device control. Each layer adds a new surface area where Claude can operate autonomously on your behalf.
The “phone-to-desktop” paradigm isn’t unique to Anthropic. OpenAI has been developing similar capabilities, and Apple Intelligence is slowly expanding what Siri can do across devices. But Dispatch is arguably the most polished implementation from a pure-play AI company so far — even if “polished” comes with a 50/50 success rate caveat.
The central question, as one tech publication put it: will everyday users pay Anthropic for what the open-source community already built for free? The answer probably depends on how fast Dispatch improves from research preview to reliable tool — and whether OpenClaw’s security issues become serious enough to push cautious users toward a paid, sandboxed alternative.
FAQ
How much does Claude Dispatch cost?
Dispatch is currently available to Claude Max subscribers ($100/month for 5x or $200/month for 20x usage). Pro subscribers ($20/month) will gain access within days. A free-tier rollout is planned for later in 2026.
Does Claude Dispatch work on Windows or Linux?
Not yet. Dispatch currently requires the Claude Mac desktop app. Anthropic hasn’t announced a timeline for Windows or Linux support.
How is Claude Dispatch different from Claude Code’s Remote Control?
Remote Control is designed for developers using Claude Code via the CLI — it lets you send coding commands from your phone to a terminal session. Dispatch is built for Cowork, Anthropic’s general productivity agent, and focuses on non-coding tasks like file management, database queries, and document generation.
Is Claude Dispatch safe to use with sensitive files?
Dispatch uses end-to-end encryption for the phone-to-desktop connection, runs code in a sandbox, and keeps all files local on your Mac. Every action requires user confirmation. That said, it’s still a research preview — Anthropic recommends not using it for mission-critical workflows yet.
What are the main alternatives to Claude Dispatch?
OpenClaw is the most prominent alternative, offering free open-source desktop AI control with broader LLM and messaging support, but with documented security concerns. OpenAI is developing similar remote agent capabilities, and Apple Intelligence handles basic cross-device task delegation within the Apple ecosystem.
You Might Also Like
- 13b Into Openai yet Microsoft Copilot Cowork Runs on Anthropic Claude
- Starnus Just hit 1 on Product Hunt and Yeah its Worth the Hype
- Lovon Just Topped Product Hunt on Valentines day and its not a Dating app
- Zenmux Just hit 1 on Product Hunt Heres why Everyones Paying Attention
- Koidex Just hit 1 on Product Hunt and its Parent Company got Acquired for 400m

Leave a comment