How To Use Claude's Massive New Upgrades

March 24, 2026 · Episode Links & Takeaways

Author

AI Daily Brief
March 24, 2026

This episode has a companion experience! Our first one!
Click through to check out our handy checklist and resources as you’re trying out all the new Claude features mentioned in the show.
Companion experiences can be found at play.aidailybrief.ai 

MAIN STORY

How to Use Claude's Massive New Upgrades

Over the past month, Anthropic has shipped a wave of new features for Claude Code and Claude Cowork — so many that it's time for a full retrospective. The lineage goes: better models unlocked new capabilities, OpenClaw showed what a fully-featured agent harness could look like, and Anthropic has been racing to bring those patterns natively into Claude ever since. The result is an always-on, context-maintaining, persistent orchestration experience where work happens even when you're not in front of your computer — and you can interact with it from wherever you are. Here's everything that's shipped, and how to use it.

Anthropic Docs Cowork Dispatch
Anthropic Docs Claude Code Remote Control
Anthropic Docs Claude Code Channels
Anthropic Docs Claude Computer Use

KEEPING UP WITH CLAUDE

Remote Control
"Not a productivity feature — a relationship shift."
The first feature in this wave: start a Claude Code terminal session on your machine, then pick it up and continue from your phone. There's no cloud relay — the session keeps running locally, with full access to your file system, MCP servers, tools, and project configs. Your phone (or any other device) is just a window into that local session. You can start a dedicated remote control server by navigating to your project directory and running claude remote-control, or switch an existing session to remote with /remote-control or /RC. Peter Levels noted it works as a drop-in for SSH. Roman Mirzoyan shipped an app update to the App Store without touching his laptop after fixing two bugs on a half-day walk. Gagan Saluja put it best: "That's not a productivity feature, that's a relationship shift. You stop thinking of it as a tool you operate and start thinking of it as something you delegate to and check in with."

Cowork Dispatch
"Your phone is the command chair. Your desktop does the heavy lifting."
Dispatch is a single persistent conversation thread in Claude Cowork that runs on your computer — message it from your phone, come back to finished work. Unlike remote control (which is a window into a session you're already running), Dispatch is an orchestrator: from one conversation on your phone, you can spawn and manage multiple Cowork tasks and Claude Code sessions running simultaneously on your desktop. Each session has its own context, file access, and connectors. When you assign a task, Claude figures out what kind of work is needed and routes it accordingly — dev tasks go to Claude Code, knowledge work goes to Cowork. You get a push notification when something's done or when Claude needs your approval. Pawel Huryn ran a day's worth of competitor research, sponsor page drafts, gap analysis, and infographic iterations across errands, a dog walk, and time at a jump arena with his kids: "The actual direction time across all of these gaps was maybe 25 minutes total. The Claude execution running in parallel was 3+ hours of work." Ethan Mollick: "Claude Cowork Dispatch covers 90% of what I was trying to use OpenClaw for, but feels far less likely to upload my entire drive to a malware site."

Claude Code Channels
"Channels is for devs who want something hackable."
Channels is Claude Code's answer to OpenClaw's Telegram integration — but it's more than that. A channel is an MCP server that pushes events into your running Claude Code session, so Claude can react to things happening while you're not at the terminal. Yes, that includes chatting with Claude Code from Telegram or Discord. But it also means CI failures, webhook payloads, monitoring alerts, or any HTTP POST can reach your running session directly. The key is using Channels to trigger Claude autonomously without you needing to be directly involved. It’s a little more advanced but this dramatically changes the use cases for Claude.

Scheduled Tasks
"Check error logs every few hours, create PRs for any actionable errors."
Scheduled tasks launched in three stages. First, Cowork got scheduled tasks at the end of February — think morning briefings, weekly spreadsheet updates, Friday team presentations. Then about a week and a half later, Claude Code desktop got local scheduled tasks (Thariq's favorite use case: check error logs every few hours and create PRs for any actionable errors). Then, a couple of weeks after that, cloud-based recurring tasks arrived — set a repo, a schedule, and a prompt, and Claude runs it via cloud infrastructure. You don't need to keep Claude Code running on your local machine. The Claude Code team has been using it to sweep open PRs, build features from approved issues, analyze CI failures overnight, and sync docs based on newly merged PRs.

Computer Use
"Not a sandbox, not a simulation — your real Jarvis."
The biggest announcement of the batch, and by far the most viral: Claude can now use your computer. Mouse, keyboard, screen, any app. The official Claude account tweet hit 40 million views and 62,000 bookmarks within 16 hours. The way it works: Claude reaches for the most precise tool first, using existing connectors to services like Slack or Google Calendar where available. When no connector exists, it falls back to controlling your browser, mouse, keyboard, and screen directly — opening files, navigating browsers, running dev tools, all with no setup required. Combined with Dispatch, this becomes especially powerful: Claude can use your computer on your behalf while you're away, creating a morning briefing on the train, making changes in your IDE and running tests, or keeping a long-running project on track. Box CEO Aaron Levie summed it up: "Computer use and the ability to write and run code on the fly are the ultimate primitives for agents to take on more and more tasks in knowledge work." Peter Gusev pointed out an underrated application: "This is a big deal for a lot of corporates who have custom crappy apps from 20 or 30 years ago" that will never natively integrate with AI.

Anthropic Blog Introducing computer use for Claude
Aadit Sheth (X) Claude behaves like a junior ops hire that lives on your laptop
Peter Gostev (X) Big deal for lots of corporates who have legacy apps with bad APIs
Aaron Levie (X) Computer use is the ultimate primitive for agents

The Rest of the Batch
"A big old checklist of things we need to try."
Running quickly through what else shipped: Cowork now has a concept of Projects (a frequently-requested missing feature). Claude Code got a code review mode where it dispatches a team of agents to hunt for bugs. The 1 million token context window is now generally available for both Claude Opus and Claude Sonnet — YC's Garry Tan: "I underestimated how powerful Opus 4.6 with 1 million tokens is. Even last year, we were absolutely hitting context limit problems." The main Claude app can now build interactive charts and diagrams. Claude for Excel and Claude for PowerPoint got upgrades, including the ability to share full context between them — pull data from spreadsheets, build tables, and update a deck without re-explaining anything. Skills are now available inside both add-ins, so teams can save standard workflows for one-click reuse. Memory and connectors are now available on the free plan. And enterprise customers got a plugin marketplace. Ethan Mollick: "The ability of the Claude team to learn from things like OpenClaw and implement features like this on a daily basis is a very strong argument that a very different software development process is possible."