Claude захватывает Slack и Figma: Anthropic строит «суперприложение»

Remember that feeling when you have forty tabs open, and you frantically copy summaries from Claude into Slack, while simultaneously trying to fix a layout in Figma? We've lived in this purgatory of "copy-paste" for years, waiting for when AI agents would finally start doing something useful. Anthropic seems to have decided that enough is enough. They didn't just give Claude "eyes" or "ears," they gave it hands that can now reach into your work tools and set things in order.

The secret to this trick lies in MCP, or Model Context Protocol. When Anthropic launched this open protocol some time ago, it seemed like a toy for developers. But today's update puts everything in its place: the company wants to make Claude a universal remote control for your digital life. Instead of waiting for each software developer to deign to write a plugin for a neural network, Anthropic created a bridge that any service can cross. And the first to cross this bridge were the heavyweights: Slack, Figma, Canva, and Asana.

It's important to understand that these aren't just "integrations" in the old sense, when a bot could only read a file. Now these are interactive applications living right inside a chat. You can compose a message to colleagues, format it, and send it to the right Slack channel without ever seeing that purple interface. You can make edits to a Canva design or check tasks in Asana without leaving the chat window. This sounds like a minor convenience, but in reality it's a tectonic shift in how we use software at all.

For a long time we've discussed AI agents as some futuristic thing that would book us airplane tickets. Reality turned out to be more mundane and useful: agents are just programs that know how to communicate efficiently with other programs. Anthropic understood that winning the war against OpenAI and Google doesn't require the highest benchmark scores. You need to become the most convenient tool in the daily workflow. While OpenAI tries to build its closed GPT store, Anthropic is betting on open infrastructure. It's the classic platform game: don't build applications yourself, build the foundation they'll run on.

There's a dose of irony in this situation. Once upon a time, Slack was marketed as the "email killer" that would centralize all communications. Now Slack itself is turning into a background service that a chatbot talks to. We're entering an era of "headless" software, where the interface of the original application becomes less important than its API and how skillfully an LLM can navigate it. If Claude becomes the place where your workday begins and ends, Anthropic wins, regardless of whose model is slightly better at solving math problems this week.

Of course, there remain questions about security and privacy. Giving a chatbot the keys to your Figma designs and Slack conversations is a serious step that requires trust. Anthropic is betting that the productivity gains will outweigh the concerns. Looking at how much time we spend jumping between windows, they may well be right. The dream of a "super app" didn't die, it just looks not like WeChat, but like a blinking cursor in a chat window that knows more about your work than you do yourself.

The main point: Anthropic is turning Claude into a full-fledged interface for all your work. If this experiment succeeds, we'll soon have no need to open a browser at all — just one chat window will suffice. Are you ready to entrust a neural network with the keys to all your work accounts?