Claude Can Draft Your Grant Proposal. Is It the Right Grant?

What arrived on October 26

Anthropic released a Claude for Nonprofits plugin on October 26, 2026. It describes skills for fundraising, marketing and organizational work, including examining a request for proposals—the funder’s application brief—and preparing a draft from the nonprofit’s own material. The announcement says the plugin is automatically installed for verified nonprofits on Team and Enterprise plans at no additional plugin cost. [1] (opens in a new tab)

This is an addition to the existing nonprofit program. The nonprofit discounts and lower Team minimum were not all introduced this week. As checked on October 27, the public pricing page lists standard nonprofit Team access at $8 per user per month, with a two-seat minimum: $16 monthly before applicable tax for two standard seats. The $3 access price is for nonprofits based in eligible low- or middle-income countries, not a general US price. Usage limits apply; Enterprise is handled through sales. [2] (opens in a new tab)

The Academy tutorial says to ask an administrator if the plugin is missing. It documents grant-fit review that distinguishes satisfied requirements, failed requirements and requirements that cannot yet be established. That is a useful design ambition. It does not mean Claude has verified your eligibility, predicted a win or replaced the person responsible for applying. [3] (opens in a new tab)

Two grants arrive. Only one deserves another conversation.

Imagine a neighborhood arts nonprofit running painting workshops for adults. It wants to add a Saturday series in a borrowed community room. Its program lead has a clear outline and an estimated materials budget, but the room booking is still being discussed. These details and both opportunities below are invented to illustrate the decision.

Grant A sounds perfect in a search result: community creativity, local access, workshop support. Read the full fictional brief, though, and it funds programs exclusively for participants under 18. The nonprofit’s proposed adult series does not fit that requirement. Rewriting “adult learners” as “community participants” would conceal the mismatch.

Grant B supports adult arts education in the right neighborhood. Its fictional brief requires a confirmed venue and an approved project budget. Those items are not yet in the folder. That is a different situation: the team has a question to resolve, not an established match.

A useful first result would show the requirement, its location in the brief, the relevant fact about the organization and what remains unresolved. A confident paragraph saying “excellent fit” is less useful than a specific missing document.

Federal opportunities illustrate why the original notice matters: Grants.gov directs applicants to the eligibility section and application instructions for each opportunity. Private funders have their own rules; the imagined grants here are not federal programs or actual offers. [4] (opens in a new tab)

Give the tool a question it can help you investigate

Start with the current funding notice, its linked guidance and amendments, the program outline, a dated budget and the relevant approved organization facts. Keep the packet small enough for someone to check. For this arts example, identifiable participant records and donor histories add little to the initial fit decision.

An instruction for the proposed exercise could be: compare this adult workshop plan with the current grant requirements; give the source section for each requirement and the document supporting our answer; mark missing information explicitly; list the questions we must resolve before deciding whether to apply. Do not change the program to make it appear eligible.

Ask the reviewer to open the source beside every decisive answer. Check the actual deadline, its time zone, accepted applicants, permitted activities, required attachments and submission method. An old search result or last year’s saved PDF should not silently become this year’s rule. If the funder’s wording remains ambiguous, the team should ask the funder rather than turn a guess into a fact.

Claude’s tutorial explains that a map of where records live is different from access to those records. Connectors or attached files provide the material; memory and cloud-task settings affect whether the map persists. Follow the organization’s data policy and administrator controls before connecting accounts. A first trial with public notices and a deliberately prepared program packet can avoid opening an entire shared drive. [3] (opens in a new tab)

Anthropic says nonprofit administrators manage access and permitted connectors, and that it does not train models on the organization’s data by default. That statement is not permission to upload every record. Decide which material is appropriate for this task, who can see the resulting work and how long it should be retained. [1] (opens in a new tab)

Eligibility is only the first decision

Suppose the arts team gets its room confirmed and its budget approved. Grant B may now be worth considering. Someone still needs to decide whether the proposed series is useful, whether staff can deliver it, and whether the reporting work is manageable alongside existing workshops.

Put the award period next to the actual program calendar. Identify who would collect the required evidence, prepare reports and keep track of restricted spending. Discuss any required contribution or spending that the grant would not cover with the person responsible for the finances. Being allowed to apply is not the same as being able to deliver the project well.

Only then move into drafting. The approved plan should determine the proposal. A sentence about reaching new neighborhoods needs support; an outcome from a different program should not drift into this one. Keep proposed attendance separate from achieved attendance, and confirmed partners separate from people you hope to approach.

The plugin’s documented proposal workflow uses application guidance and existing organization material, with review of the cited figures. Treat the first draft as something to inspect. Preserve unanswered questions for staff to resolve; do not fill them with plausible numbers. [3] (opens in a new tab)

Keep submission with the responsible person: check the finished application against the current instructions and use the required channel. The documented drafting features do not establish that the plugin can submit through any particular grant portal.

A useful first trial can end with fewer applications

Before connecting the tool to daily work, prepare a small evaluation using an opportunity the team already understands, a clear mismatch and a case with missing information. Write down the expected reasons first. Check whether the proposed workflow surfaces those reasons without inventing facts or quietly overlooking a restriction.

Measure time from opening the documents to a reviewed decision, including corrections and source checks. Record missed requirements and unsupported statements as well as time saved. A shorter draft-writing step does not help if it creates a longer cleanup. A few examples can reveal problems; they cannot establish a grant-winning advantage.

If this becomes repeated work, a modest internal workspace could retain each opportunity’s source, checked date, deadline, unresolved question and responsible staff member. Link the reviewed decision to the eventual proposal and award report. That is a possible custom-software project, not something every nonprofit needs before trying the plugin.

The public website can support the same discipline: keep current program descriptions and published impact figures accurate so staff and prospective supporters are not reading contradictory accounts. Keep private applications and decision notes behind appropriate access controls.

The new tools are interesting because they reach beyond an empty text box. For a small team, the first worthwhile outcome may be a clear decision to skip Grant A, ask two questions about Grant B, and get back to preparing the workshop. Whether Claude makes that process better is something to test against your own work.