{"id":18874,"date":"2024-11-25T19:23:56","date_gmt":"2024-11-25T19:23:56","guid":{"rendered":"https:\/\/gpt.m2mbeta.com\/?p=18874"},"modified":"2024-11-25T19:23:56","modified_gmt":"2024-11-25T19:23:56","slug":"anthropic-proposes-a-new-way-to-connect-data-to-ai-chatbots","status":"publish","type":"post","link":"https:\/\/gpt.m2mbeta.com\/?p=18874","title":{"rendered":"Anthropic proposes a new way to connect data to AI chatbots"},"content":{"rendered":"
\n

Anthropic is proposing a new standard for connecting AI assistants to the systems where data resides. <\/p>\n

Called the Model Context Protocol, or MCP for short, Anthropic says the standard, which it open sourced today, could help AI models produce better, more relevant responses to queries. <\/p>\n

MCP lets models \u2014 any models, not just Anthropic\u2019s \u2014 draw data from sources like business tools and software to complete tasks, as well as from content repositories and app development environments.<\/p>\n

\u201cAs AI assistants gain mainstream adoption, the industry has invested heavily in model capabilities, achieving rapid advances in reasoning and quality,\u201d Anthropic wrote in a blog post<\/a>. \u201cYet even the most sophisticated models are constrained by their isolation from data \u2014 trapped behind information silos and legacy systems. Every new data source requires its own custom implementation, making truly connected systems difficult to scale.\u201d<\/p>\n

MCP ostensibly solves this problem through a protocol that enables developers to build two-way connections between data sources and AI-powered applications (e.g. chatbots). Developers can expose data through \u201cMCP servers\u201d and build \u201cMCP clients\u201d \u2014 for instance, apps and workflows \u2014 that connect to those servers on command.<\/p>\n

\n

Here’s a quick demo using the Claude desktop app, where we’ve configured MCP:<\/p>\n

Watch Claude connect directly to GitHub, create a new repo, and make a PR through a simple MCP integration.<\/p>\n

Once MCP was set up in Claude desktop, building this integration took less than an hour. pic.twitter.com\/xseX89Z2PD<\/a><\/p>\n

\u2014 Alex Albert (@alexalbert__) November 25, 2024<\/a><\/p><\/blockquote>\n

Anthropic says that companies including Block and Apollo have already integrated MCP into their systems, while dev tooling firms including Replit, Codeium, and Sourcegraph are adding MCP support to their platforms. <\/p>\n

\u201cInstead of maintaining separate connectors for each data source, developers can now build against a standard protocol,\u201d Anthropic wrote. \u201cAs the ecosystem matures, AI systems will maintain context as they move between different tools and data sets, replacing today\u2019s fragmented integrations with a more sustainable architecture.\u201d<\/p>\n

Developers can start building with MCP connectors now, and subscribers to Anthropic\u2019s Claude Enterprise<\/a> plan can connect the company\u2019s Claude chatbot to their internal systems via MCP servers. Anthropic has shared pre-built MCP servers for enterprise systems like Google Drive, Slack, and GitHub, and says that it\u2019ll soon provide toolkits for deploying production MCP servers that can serve entire organizations.<\/p>\n

\u201cWe\u2019re committed to building MCP as a collaborative, open-source project and ecosystem,\u201d Anthropic wrote. \u201cWe invite [developers] to build the future of context-aware AI together.\u201d<\/p>\n

MCP sounds like a good idea in theory. But it\u2019s far from clear that it\u2019ll gain much traction, particularly among rivals like OpenAI, which would surely prefer that customers and ecosystem partners use their<\/em> data-connecting approaches and specifications.<\/p>\n

In fact, OpenAI recently<\/a> brought a data-connecting feature to ChatGPT, its AI-powered chatbot platform, that lets ChatGPT read code in dev-focused coding apps \u2014 similar to the use cases MCP drives. OpenAI has said that it plans to bring the capability, called Work with Apps, to other types of apps in the future, but it\u2019s pursuing implementations with close partners rather than open sourcing the underlying tech.<\/p>\n

It also remains to be seen whether MCP is as beneficial and performant as Anthropic claims that it is. The company says, for example, that MCP can enable an AI bot to \u201cbetter retrieve relevant information to further understand the context around a coding task,\u201d but the company offers no benchmarks to back up this assertion. <\/p>\n<\/div>\n