Anthropic just dropped Channels for Claude Code. It's the feature that officially makes OpenClaw's biggest advantage irrelevant. Claude can now receive messages from Telegram, Discord, iMessage, Slack, and webhooks directly into a running session. The always-on AI agent that OpenClaw promised? Anthropic just built it natively.

Here's what happened. Claude Code shipped a new flag called --channels as a research preview. It lets MCP servers push messages into your Claude Code session while you're away. CI results, Slack messages, monitoring alerts, chat messages from any platform. Claude sees them and reacts.

That was OpenClaw's entire pitch. A persistent AI agent wired into your messaging apps that runs 24/7 and responds to you wherever you are. OpenClaw had 15+ platform integrations. Claude had zero. That gap just closed.

But Channels isn't even the whole story. Two days ago Anthropic shipped Dispatch for Cowork. That lets you text Claude from your phone and it runs tasks on your desktop while you're gone. Before that they shipped voice mode for Claude Code. Before that, /loop for recurring background tasks. Before that, scheduled tasks in Cowork.

Every single one of these features maps directly to something that made OpenClaw the fastest growing open source project in GitHub history. Anthropic has been checking off the list one by one.

And here's the part that really matters. Anthropic did it without the security disasters. OpenClaw had a critical vulnerability earlier this year with a CVSS score of 8.8. Researchers found over 50,000 exposed instances. A security audit of OpenClaw's community skills registry found roughly 12% were malicious. The creator himself admitted to shipping code he doesn't read.

Claude Code runs in a sandbox. Permissions are explicit. Actions require approval. Enterprise admins control what's enabled. No community skill registry full of data exfiltration scripts.

Because Anthropic just proved that the best way to beat an open source viral sensation is to build the same thing better and ship it inside the product people already pay for.

OpenClaw exploded because it solved a real problem. People wanted an AI that doesn't stop working when you close the tab. An AI that lives in your messages and remembers what you told it last week. An AI that runs in the background while you sleep.

Anthropic watched all of that happen and said we can do this natively. Dispatch gives you the phone-to-desktop remote control. Channels gives you the messaging integration. /loop gives you the recurring tasks. Cowork gives you the file access and browser control. All inside a product with real security, real enterprise controls, and Opus-level reasoning.

OpenClaw still has 325,000 GitHub stars and a passionate community. It's not going away tomorrow. But the moat just got a lot shallower. Every week Anthropic ships another feature that removes another reason to run a third party agent on your machine.

Look at what Anthropic has shipped in the last 10 days. Dispatch. Channels. Voice mode. /loop. Scheduled tasks. Plugin marketplace. Enterprise connectors. Windows support. 1M context on all paid plans. A 2x usage promotion. Oh, and Katy Perry just posted herself signing up for Claude Pro. That's the kind of month Anthropic is having.

Most companies would slow down while fighting the Pentagon in court and watching $285 billion get wiped off SaaS stocks because of their product. Anthropic sped up.

OpenClaw proved millions of people want always-on AI agents. It also proved the open source approach has real limits. Security holes. Malicious skills. No enterprise controls. Anthropic saw the demand signal and decided to own it before someone else did it properly. Every feature they've shipped this month is a direct answer to a gap that OpenClaw exposed.

Claude isn't trying to be the smartest AI anymore. It's trying to be the one you never turn off.

Cursor dropped Composer 2, their own in-house coding model. It beats Claude Opus 4.6 on several benchmarks and is 86% cheaper than their previous model. It's built specifically for long-horizon agentic coding inside Cursor. Bloomberg framed it as Cursor trying to break free from depending on Anthropic and OpenAI for its intelligence. The company now has over 1 million daily users, $2 billion in annual revenue, and is in talks to raise at a $50 billion valuation. The AI coding wars just got another player with their own model.

Honestly this is all of us right now. Anthropic has shipped more features in March than most companies ship in a quarter.

Anthropic shipped Channels for Claude Code. Telegram, Discord, Slack, iMessage can now push messages directly into a running session. Combined with Dispatch, voice mode, and /loop, Claude now does everything OpenClaw does but inside a secure paid product with Opus-level reasoning. Cursor dropped their own model to stop depending on Anthropic. One company is killing the competition. The other is trying to survive it.

Stay building. 🤖