Funding Landscape Supplies the Data; Gemini Supplies the Interface
Gemini is useful for turning an incomplete description of an organization into search angles, comparing eligibility language, building a deadline plan, and explaining an unfamiliar program. It is not a maintained funding database. Programs change, deadlines pass, notices are amended, and a model can produce a specific answer that sounds more current than it is. The Funding Landscape connection closes that gap. Gemini sends structured searches to Funding Landscape and receives maintained records with source links. That is more reliable and repeatable than asking an assistant to rediscover the funding universe one web page at a time. You still open the primary source before acting, but the discovery and qualification queue begins with the product's live corpus rather than model memory. This guide covers Gemini CLI because its official documentation supports remote Model Context Protocol servers. It does not claim that every Gemini web or mobile product has the same connector interface. If you use another compatible assistant, the broader AI grant-search guide explains the same Funding Landscape workflow.
What the Connection Does
Model Context Protocol, or MCP, lets an assistant call an external tool. The assistant decides when a funding search would help, sends structured search terms and filters to Funding Landscape, and receives real records with funder, deadline, eligibility context, and a source URL. Gemini still explains and compares the results, but it does not need to invent the underlying opportunity. The distinction matters. A database result is not automatically a recommendation. You still need to confirm applicant type, geography, deadline, award terms, source documents, and whether the program can pay the cost you have. The connector improves factual grounding and coverage. It does not replace due diligence or promise eligibility. Funding Landscape's free account currently includes 10 MCP search calls per month, up to 100 results per call, with newly discovered listings appearing after a delay. Paid plans raise the monthly call limit and provide same-day listings. Those are Funding Landscape limits, separate from any Gemini API or Google account limits.
🔍 Search related opportunities now
Connect Gemini CLI to Funding Landscape
Open the Funding Landscape AI setup page in one tab. Install and sign in to Gemini CLI by following Google's official Gemini CLI documentation, then run this command in your terminal: gemini mcp add --transport http --scope user funding-landscape https://fundinglandscape.com/api/mcp The official Gemini CLI MCP server guide documents the gemini mcp add command, remote HTTP transport, server listing, and OAuth. The user scope makes the server available across your Gemini CLI projects. Omit --scope user if you want Gemini's default project-level configuration instead. Restart Gemini CLI after adding the server, then run /mcp list. Funding Landscape should appear as connected and its tools should be visible. The server permits limited anonymous use, so simply connecting may not open a browser. To link a Funding Landscape account, run /mcp auth funding-landscape. Gemini CLI should discover Funding Landscape's published authorization metadata and open the browser flow. You can also ask Gemini to use the Funding Landscape sign_in tool; that tool deliberately returns the authorization challenge when no account is linked. Sign in or create an account in the browser, authorize the connection, and return to the same CLI session. After authorization, ask: "Check my Funding Landscape account status." A successful response should identify the linked account's access level and current usage. You do not need to copy a secret token into a settings file. Do not paste credentials, private proposal material, or confidential applicant data into a public issue or shared configuration.
A Better First Search
Do not begin with "find grants for my nonprofit." Give Gemini a compact eligibility profile and ask it to search in stages. For example: "We are a 501(c)(3) youth arts organization in Ohio. We run after-school programs for ages 11 to 15 and need teaching-artist wages, supplies, and transportation. Our annual budget is $350,000. Search current opportunities we could apply to, separate operating support from project grants, and cite the source record for every result." That prompt supplies applicant type, geography, population, activity, cost, and scale. Those facts help Gemini choose better search terms and reject obvious mismatches. Ask it to return a table with program, funder, applicant type, geography, deadline, amount, match, allowed cost, and source URL. Require "unknown" when a field is absent. An honest blank is safer than a guessed deadline. Then use a second prompt: "For the five strongest results, explain one reason we fit, one unresolved eligibility question, and the next verification step. Do not infer eligibility that the source does not state." This turns a list into a qualification queue.
Search Broadly Without Burning Every Call
One well-formed MCP search can return up to 100 results, so do not spend a call on every synonym. Start with a broad but coherent problem, inspect the result language, and refine only when the top records reveal a useful lane. Plus and Pro accounts can send two to five related query variations as one paid batch call, then receive a deduplicated merged result set. For a rural clinic, the first batch might separate rural health access, telehealth, workforce, and facility improvement. For a manufacturer, separate research and development, workforce training, energy upgrades, export support, and public procurement. For an arts nonprofit, separate education, presenting, capital, and operating support. Ask Gemini to remember exclusions within the conversation: no loans, no individual fellowships, no opportunities restricted to governments, or no deadlines inside 14 days. Keep important filters in the search prompt too. Conversational memory is helpful, not a substitute for explicit criteria. When you want to browse directly, use Funding Landscape search. When you want the assistant to combine your organization context with current Funding Landscape records, use the AI connection. Both interfaces draw on the same maintained opportunity corpus.
Verify Every Shortlisted Opportunity
For each serious result, open the primary source. Confirm that the notice is still open, the deadline includes the correct time zone, your legal applicant type is eligible, your location is allowed, and your proposed cost is permitted. Read amendments and attachments, not only the summary page. Ask Gemini to make a verification checklist, but supply the source text or let it inspect the source where browsing is available. A useful prompt is: "Compare this official notice with the database record. List any difference in deadline, award size, eligibility, match, or application route. Quote only short phrases and link the official source." If the record and source disagree, the source controls. Report the discrepancy so the maintained data can be corrected. Do not rationalize the mismatch because the opportunity otherwise looks attractive. The federal funding search guide maps the official systems and what each one can verify. For federal opportunities, check Grants.gov, SAM.gov, or the issuing agency. For state and local programs, use the official funding or procurement portal. For foundations, use the foundation's own current guidelines. A directory or article can help you discover a program, but it should not be the final authority for an application.
Turn Search Results Into a Weekly Decision System
Have Gemini maintain a small pipeline with four statuses: verify, pursue, watch, and reject. Every record should include the source, deadline, owner, next action, and rejection reason. Rejection reasons are valuable data. After a month, you may learn that your searches return too many government-only programs, that match requirements are the main constraint, or that you need a fiscal sponsor. Use saved searches and alerts for lanes that repeatedly fit. A weekly prompt can ask Gemini to search for new records, compare them with the pipeline, and highlight deadline or eligibility changes. Do not ask it to write a full proposal before deciding whether the opportunity is real and strategically useful. The best division of labor is simple: the live source provides the opportunity facts, Gemini helps reason across them, and a human owns the final eligibility and application decision.
Troubleshooting the Connection
If Gemini shows the server as disconnected, run /mcp list and inspect the server status. The detailed Gemini CLI MCP server guide documents OAuth discovery, browser authorization, and remote Streamable HTTP connections. Run /mcp auth funding-landscape to start or restart account linking. If a browser opens but the connection does not complete, confirm that the callback returned to the same Gemini CLI session, then restart the CLI and list the servers again. If tools are connected but Gemini does not use them, ask explicitly: "Use the Funding Landscape tools to search current records." Do not confuse anonymous access with a linked account. The server can connect and expose search tools before authorization. A 401 returned specifically by the sign_in tool is the protocol signal that starts OAuth; repeated failures after the browser flow are an error. Do not work around them by placing an account password or private token in a prompt. If the connector remains unavailable, search the database directly and paste only the non-confidential result details you want Gemini to compare. You can still use the reasoning workflow without giving the model permission to invent current programs.