A new way to access nonprofit data: Candid MCP connector for Claude 

The AI wave has officially reached nonprofits. Recent surveys indicate that over 85% of nonprofits are exploring AI tools, with up to a third making proactive efforts to integrate them across their organizations. As with many in the commercial space, social sector groups are drawn to AI’s potential to boost productivity for their teams. More than 60% of nonprofits surveyed said they were specifically interested in AI’s potential to support grant writing and optimize fundraising efforts. At the same time, we know that nonprofits and funders are concerned about the relevance and trustworthiness of information, especially when it’s sourced via AI tools. 

As our CEO, Ann Mei Chang, wrote last month, Candid is more committed than ever to centering our users in the technology and tools we build. To get people the information they need to do good, we need to meet users where they are. And that means meeting them in their actual workflow tools—including AI assistants. 

Announcing Candid via Claude for Nonprofits 

This week, Candid launched a new way to access our comprehensive data on nonprofits and funders: the Candid MCP (Model Context Protocol) connector for Claude, Anthropic’s AI assistant. Candid is excited to be a partner for Claude for Nonprofits and to be listed in the Claude Connectors Directory. Announced on GivingTuesday, Claude for Nonprofits includes discounted Claude subscriptions for eligible organizations, along with tools like the Candid MCP connector and support tailored to the social sector.  

The Candid MCP connector offers a direct, secure pipeline from Claude to Candid’s data and search functionality. It provides access to our nonprofit data, funder information, and sector knowledge—all through Claude’s chatbot interface.  

Available to all registered Candid users, including those on our free tier, the Candid MCP is a level-up from generic AI: Whereas Claude alone relies on often outdated training data or unstructured and unvalidated web search, Claude plus the Candid MCP means you can quickly get authoritative answers about nonprofits and funders from verified primary sources, updated in real time. 

What is an MCP, exactly? 

Model Context Protocol is one of the biggest new concepts to emerge in the AI space. It’s a standard way for large language models (LLM)s or AI assistants to securely access and interact with data, files, tools, or software, either in the cloud or on your computer, making AI more personalized and useful. The Candid MCP is available through the Claude Connectors Directory, meaning you don’t have to configure an API or write any code. All you have to do is click “Connect” and log in with your Candid account, and Claude instantly knows when and how to use it. You can ask questions in the Claude app like “What funders are supporting early childhood education in Northern California?” and get responses back that are sourced from Candid’s free-tier data, plus links to Candid profiles for clear citations and deeper research.  

Under the hood, Claude and the Candid MCP are ‘talking’ to each other to pass on your question or request, identify keywords and taxonomy, apply the correct filters, query the database, and link to Candid profiles, then translate it back into natural language in your chat. You can see full documentation of the tools in our user guide.  

What can you do with the Candid MCP connector? 

The power of Claude plus the Candid MCP is that it can be adapted and customized to your needs—including combining it with your other tools and data. But here are some examples of what you can do by asking questions in Claude: 

• Find organizations: Search for nonprofits and foundations by name, mission, location, leadership demographics, or area of work. 
   
  Example: “Help me find nonprofits working on food access in Seattle that are highly transparent” 
     
• Discover connections: Identify funders and nonprofits working in specific issue areas or geographic regions—especially powerful when paired with a Candid subscription for further research. 
   
  Example: “What are some funders in the southern U.S. that support environmental sustainability programs?”   
    
• Access expert knowledge: Search Candid’s library of research reports, training materials, blog posts, and curated news. 
   
  Example: “How do I write an effective grant proposal?”  

Feedback welcome 

As an emerging technology, AI tools like the Candid MCP connector are inherently experimental; we know we’re just at the beginning of understanding what’s possible. We want to improve and expand it based on your feedback. What queries work? Where are the gaps? What other Candid data would help? How does this fit your actual workflow? 

Please tell us at onlinelibrarian@candid.org, and check out our user guide for more information. We’re excited to shape the future of AI tools at Candid together.  

If you’re looking for your own systematic integration with Candid’s data, explore options and connect with one of our experts.