Anthropic Dispatch feature enabling Claude AI to perform desktop tasks autonomously from mobile instructions

Anthropic Ships Dispatch for Claude Cowork

Text a Task from Your Phone, Let Claude Handle the Rest on Your Desktop – Files, Docs, Folders, All in the Background

In a quiet but game-changing update, Anthropic bridges mobile and desktop AI like never before – turning Claude into a truly persistent, autonomous coworker that works while you walk away.

By VFuture Media Staff | March 18, 2026

Anthropic just dropped one of the most practical agentic AI features of 2026 so far: Dispatch for Claude Cowork.

Announced March 17 as a research preview, Dispatch lets users kick off complex, multi-step tasks from their phone — then step away while Claude executes them on the desktop Mac (with Windows support in the works). No babysitting required. No constant monitoring. Just describe the job via text message, and come back to organized folders, built documents, synthesized reports, or cleaned-up data ready and waiting.

The setup is refreshingly straightforward. Update the Claude desktop app, enable Dispatch in the Cowork tab, scan a QR code from your phone (iOS or Android Claude app), and pair the devices. Once connected, the persistent Cowork session lives on your computer in a secure, sandboxed environment — accessing only the files and folders you explicitly grant permission for. Nothing uploads to Anthropic’s servers; your data stays local.

This isn’t just remote control — it’s asynchronous agentic execution. You text something like:

  • “Organize my Q1 project screenshots into dated folders, extract key metrics into a new Google Sheet, and draft a one-page summary email.”
  • “Scan my Downloads folder for all PDFs from last month, rename them with client names from the filename, move to Archives/2026, and flag any duplicates.”
  • “Pull the latest sales CSV from Dropbox, merge with the CRM export in my Documents folder, run basic analysis, and create a formatted report in Word.”

Claude plans the steps (showing you for approval if needed), breaks them into sub-tasks, uses its built-in tools to read/write files, launch apps, browse if permitted, and iterate until done. You can check progress from your phone in real time or just close the app and go about your day. When you return to your Mac, the results are there — files restructured, docs generated, folders tidied.

Early users and reviewers are already calling it “90% of what people wanted from OpenClaw, but without the security roulette.” Unlike more experimental always-on agents that risk broad system access or cloud uploads, Dispatch inherits Cowork’s safety-first design: sandboxed VM, default-deny networking, explicit permissions, and transparent step-by-step reasoning before any action touches your files.

Availability rolled out first to Claude Max subscribers ($100–200/month tier), with Anthropic promising rollout to Pro users ($20/month) within days. It’s currently Mac-focused (Claude Desktop app required, machine must stay awake), but the architecture suggests broader platform support is coming.

Why this matters now: Agentic AI has been stuck between two worlds — flashy demos that impress on stage but fizzle in real workflows, or powerful but intimidating tools like Claude Code that demand terminal fluency. Cowork already made agentic capabilities accessible to non-devs. Dispatch removes the last major friction: you no longer need to be glued to your computer to leverage it.

For knowledge workers juggling meetings, travel, family, or just wanting to reclaim evenings, this is the “set it and forget it” moment many have waited for. Text a task during your commute, hit the gym, pick up kids — come back to finished work. It’s not magic; it’s meticulous engineering making AI feel like a true coworker.

Anthropic’s safety obsession — the same philosophy that built Constitutional AI — shines here. By keeping execution local, sandboxed, and permissioned, Dispatch sidesteps the biggest risks plaguing less constrained agents. It’s production-grade agentics for the rest of us.

As the agent wars heat up post-NVIDIA GTC, Anthropic isn’t trying to out-flash anyone. They’re quietly shipping tools that actually ship value — today, on your existing hardware, without rewriting your security model.

Dispatch for Claude Cowork isn’t just a feature. It’s proof that the future of work isn’t about AI replacing you — it’s about AI finally working for you, even when you’re not watching.

VFuture Media will keep tracking the agentic AI wave as it moves from hype to everyday utility. From Dispatch to the next frontier of physical and digital agents, the tools that quietly change how we work are arriving fast.

Post navigation

Leave a Comment

Leave a Reply

Your email address will not be published. Required fields are marked *