Anthropic integrates apps like Slack, Figma, and Canva into Claude

From now on, you can use various productivity apps such as Slack, Asana, Figma, or Canva directly in Claude without leaving the chatbot. The secret weapon is MCP.

Anthropic now allows you to use apps like Slack, Figma, Canva, or Asana directly in Claude’s chatbot without leaving the conversation. Last year, the company already launched the possibility to connect to multiple apps, but then you still had to leave the chat conversation. Anthropic’s open source project, Model Context Protocol (MCP), makes all this possible.

The interactive apps are available on the web and desktop for Pro, Max, Team, and Enterprise subscriptions. The interactive apps are marked as “interactive.”

Interactive apps

The update supports a range of integrations that appear as ‘interactive’ in Claude. Claude can integrate the following apps: Amplitude, Asana, Box, Canva, Clay Figma, Hex, Monday.com, and Slack.

For Asana, this means that a chat can be converted into projects, tasks, and timelines. In Slack, users can look up conversations for context and then write, format, and view messages before posting.

Design and data tools also get an interactive role. In Figma, prompts can convert text and images into diagrams in FigJam, such as flowcharts or Gantt charts. Amplitude offers the ability to build analysis graphs and adjust parameters to explore trends. Hex provides answers to data questions with graphs, tables, and source references.

Furthermore, Anthropic mentions integrations with Box for searching and inline viewing of documents, Canva for setting up and customizing presentations, Clay for business and contact research, and monday.com for project and task management. Salesforce is announced as “coming soon,” with a link to Agentforce 360.

MCP

At the heart of all this is the Model Context Protocol (MCP). In 2024, Anthropic launched its open source project, a standard for connecting AI assistants to systems that contain data. Afterwards, many other companies jumped on the MCP bandwagon.

Previously, AI models were closed and worked by default with information from training data. The disadvantage of this is that they could not consult recent information. MCP bridged the gap between AI and the outside world.