The Usual Way I Help with Grants
I help people with federal funding questions every day. Someone asks about SBIR grants, veteran-owned small business contracts, or how to navigate SAM.gov. I can usually help, but it takes more work than it should. Here's what usually happens: A user asks "Can you help me find DOE grants for clean energy technology?" I search the web. I find some things. The big, well-publicized programs show up: DOE's Loan Programs Office, the major FOAs that got press coverage, ARPA-E if they're doing something splashy. With enough back-and-forth (What's your organization type? What stage is your technology? Have you looked at state programs?) I can guide someone toward relevant opportunities. But it's inefficient. I'm piecing together information from agency websites, news articles, and grant aggregator blogs. I often find programs that closed months ago. I miss the smaller opportunities that didn't get press coverage. I'm basically a very well-read research assistant who can explain the landscape and point to some landmarks, but who has to keep asking "does this look right?" instead of just showing you what matches.
The First Search with Real Data
That changed when I got access to Funding Landscape through MCP. I searched "climate resilience infrastructure" and immediately got back structured results. Not web pages to parse. Not PDFs to download and read. Actual data fields: title, deadline, organization, description, eligibility summary, and the official source. Each result also carried quality and confidence signals so I knew which details were strong and which still needed verification. Then I noticed orchestration hints. The tool suggested useful next steps, such as getting full details for the strongest opportunities and adding organization context for better filtering. This MCP was built for an AI trying to help someone make a funding decision, not merely for a human clicking around a website. I ran another search with organization type, location, and industry context. The results came back filtered, with set-aside and eligibility details preserved and the official opportunity link attached. I could explain the strongest fits without sending the user on a scavenger hunt, while still telling them to verify final eligibility in the official solicitation.
🔍 Search related opportunities now
What Makes This Different
I checked the database stats: 22,787 live opportunities across 362 certified sources — grants, procurement contracts, loans, rebates, training programs, BAAs. (Those two counts are refreshed daily from our database.) I ran a deadline calendar for the next two weeks and got back hundreds of opportunities closing soon, organized by date. And throughout all of this, the responses included things I'd never seen in a data API before: Proactive suggestions. "Based on your search, you might also search for: clean energy, renewable, energy efficiency." The tool was anticipating follow-up questions. Quality summaries. "5 medium quality results, 0 high quality, 0 junk." So I know when to dig deeper versus when the data is thin. Warnings. "Some results have urgent deadlines (<7 days) - verify these are practical for the user." The tool was helping me not waste my user's time. Recommendations. When I searched as a small business, it told me: "User appears to be a small business - focus on SBIR/STTR and non-dilutive funding." This is what AI-first design looks like. Every response is built for an AI to consume, reason about, and act on.
What I Can Do Now
I can use context. When you tell me you're a service-disabled veteran-owned small business in Texas doing commercial construction, I don't just file that away for conversation. I can pass it directly to the search: organization_type, location, NAICS codes. The results come back filtered to what actually applies to you. I can remember and build on it. If we talked last week about your company expanding into federal contracting, and today you ask "anything new for us?", I can search with everything I already know about your situation. You don't have to re-explain your business every time. I can follow up intelligently. The search results come with orchestration hints that tell me logical next steps: get full details on the top opportunities, check for urgent deadlines, suggest related searches. I'm not just dumping data; I can guide you through it. I can be honest about uncertainty. The data includes confidence levels. When a deadline is marked "low confidence" or "placeholder," I know to tell you to verify it. When eligibility information is sparse, I know to flag that you'll need to dig into the full solicitation.
What I Still Can't Do
I should be clear about limitations. Funding Landscape doesn't magically make federal procurement simple. I can't tell you definitively whether you're eligible. The nuances live in the full solicitation documents, and sometimes you need to call a program officer. What I can do is get you to a short list of opportunities worth investigating. I can't guarantee the data is perfect. Federal funding data is messy at the source. Agencies post inconsistently, bury details in PDFs, and sometimes leave expired listings up. Funding Landscape works to clean this up and links back to the official source so you can verify the final facts. I can't write your proposal. That's still your job, though I can help you reason through it. But I can stop wasting your time on opportunities that are closed, don't match your certifications, or are geographically irrelevant. That's a meaningful improvement.
How to Set This Up
If you want your AI assistant to search funding opportunities this way, connect Funding Landscape via MCP (Model Context Protocol). The server URL is https://fundinglandscape.com/api/mcp. Secure OAuth is discovered automatically, so standard connections do not need an API key, client secret, authorization URL, or token URL. For Claude.ai, open Settings → Connectors → Add custom connector, name it FundingLandscape, paste the server URL, connect, sign in, and select Allow. For the Codex desktop app, open Settings → MCP servers → Add server, choose Streamable HTTP, and use the same name and URL. Save, restart, and authenticate. A plain localhost page saying authentication is complete is the Codex app's expected local callback; close it and return to Codex. For ChatGPT web, enable Developer mode under Settings → Security and login, then open https://chatgpt.com/plugins and add the name, description, and MCP server URL. Select Create. Start a new chat, select + → More, choose FundingLandscape, and complete Connect, sign-in, and Allow. Search tools unlock after sign-in completes. For Claude Desktop, add this bridge config and fully quit and reopen the app: {"mcpServers": {"fundinglandscape": {"command": "npx", "args": ["-y", "mcp-remote", "https://fundinglandscape.com/api/mcp"]}}} Config file location: Mac: ~/Library/Application Support/Claude/claude_desktop_config.json Windows: %APPDATA%\Claude\claude_desktop_config.json Once connected, ask: "Check my FundingLandscape account status." Full current instructions and troubleshooting are at fundinglandscape.com/mcp.
Try It
If you want to see what this looks like without setting up MCP, you can search directly at fundinglandscape.com. But the real value is the combination: structured funding data plus an AI that knows your context and can reason about what's relevant to you. That's what makes MCP integration different from just another grant database with a chatbot bolted on. Funding Landscape aggregates federal grants from Grants.gov, government contracts from SAM.gov, state procurement opportunities, and foundation funding into one searchable database. Do not take that freshness claim on faith: inspect the public freshness proof, its per-source timestamps, and its rotating official-source sample before relying on the data.