CARE’s fundraisers and program teams share skills and save hours each week with Claude Enterprise

Small teams stretched across too many priorities as funding shrank

CARE’s work rests on small teams in fundraising, finance, legal, communications and program design, and each team carries many priorities at once. Staff had plenty of good ideas and the skills to carry them out, but a team stretched across that many priorities cannot give every task the attention it deserves. Fragmented systems across a global organization make this harder. Colleagues often did the same kind of work from scratch when past experience could have informed it. In fundraising, for example, each of the roughly 20 fundraisers on the Foundation Partnerships team researched prospects and prepared donor communications independently, with limited shared processes or templates to build on.

The pressure grew as funding across the humanitarian sector contracted over the past 24 months. Every function, from back-office support to partnership development to program delivery, already ran at or beyond its limits, and each was asked to do even more with less.

A pilot that staff asked to keep

CARE explored several AI tools, and Claude Enterprise kept coming up as the one with potential across many business functions. CARE ran a nonprofit pilot from January to July with a limited number of staff in several business units who self-selected to join the pilot. Staff who took part were thrilled to have access to Claude Enterprise, and word of mouth spread quickly, leading to demand for access across CARE that far exceeded the number of seats in the pilot. Jerry Tuan, CARE’s Chief Information Officer, said, “Where Claude has really stood out is in its adaptability and being able to right-size to the task in front of a wide range of teams.”

Shared skills and new capacity, starting with fundraising

Staff in data and evidence, finance, legal, communications, resource mobilization, and program design and delivery now use Claude, and several teams have packaged their own methods as skills, reusable sets of instructions that Claude loads when a task calls for them, that colleagues can reuse. Fundraising shows most clearly how that works.

The Foundation Partnerships team uses Claude for prospect research, which covers vetting, funder briefs, peer checks and capacity statements, and for first drafts of proposals and reports, copyediting and analysis of grant data. A shared set of skills now applies the CARE Style Guide, produces initial drafts in CARE’s voice for human review, tests whether a draft makes the case for “Why CARE?”, cross-checks Salesforce records and compares proposals with those of peer organizations.

Beyond headquarters, staff in one country office connected several skills into a single workflow that carries a proposal through its whole life, from finding funding opportunities to submission. The workflow runs in seven stages. It finds opportunities through a daily tracker that scans, filters and shortlists funding calls, qualifies each against the office’s eligibility and the donor’s rules, and plans the timeline, owners and reviews. It then analyzes the call alongside past performance, donor templates and evidence, designs program options from that validated input while drawing on external evidence libraries so each design starts from what has worked elsewhere, and builds the full proposal package, from internal decision papers to donor-facing narratives and budgets. It ends at a human decision gate, where a person makes every real call on whether to pursue, submit and accept. Finance, partnerships, compliance and safeguarding run in parallel throughout. The team’s guiding principle is that AI drafts and checks while humans decide.

A small team in the country office built the workflow from the bottom up. They began by experimenting with AI on live proposals, built several separate tools and then connected them into one workflow, rather than waiting for a central technical team to hand them a solution. It is a working prototype, and because it was built in the field rather than handed down from headquarters, other country offices are now asking to use it, an early signal of success.

A program colleague preparing a climate funding proposal needed to identify the exact areas at greatest risk. He combined 31 publicly available data sets into an interactive dashboard, built with Claude Cowork in about 70 hours, that maps flood-prone areas with other risk indicators layered on top. CARE historically would only have been able to achieve this by hiring a consultant for that work, which he estimated at several months and about $100,000, effectively rendering it impossible to pursue. The presentation and interactive file now serve as a template that other teams can reuse with different data sets.

Other teams have gained the same kind of capacity. Consolidating budget data from CARE’s member countries used to take weeks. Claude now returns an executive dashboard and a list of flagged countries in minutes, a finance analyst vets the results, and the full analysis takes under a day. The team that monitors participant feedback connected Claude to its database and scanned 25,000 responses for possible safety and safeguarding concerns, a review it could not run efficiently before.

CARE’s AI usage guidelines keep donor, funder and program information out of public AI tools unless the digital team has approved a tool for confidential data. Tuan described the principle behind them. “Claude drafts and accelerates, but a person owns the judgment call and the final sign-off, no matter what the process is.”

More time for funders, participants and the work itself

The Foundation Partnerships team reports using Claude every day and saving 3 to 5 hours a week per person, which adds up to roughly 60 to 100 hours weekly across a team of about 20. Fundraisers now track shifts in donor priorities, analyze foundation grant data and IRS Form 990 filings, and identify new prospective partners faster, which leaves more energy for building relationships with funders. Faster editing leaves more time for in-depth technical writing and for submitting more proposals and concepts.

Relationship managers also work with program colleagues to cross-reference LinkedIn contacts with foundation prospects to plan outreach to current and new foundations. The Foundation Partnerships team began using Claude this way about a month ago, so it is too early to tell whether the work has led to funding or major relationship developments.

CARE advises other nonprofits to put licensing and governance in place early, to start with concrete, high-impact and low-risk administrative uses, and to have training and support ready from the start. “Moving together and being open to lots of experiences and concerns has been just as critical to our momentum as having access to the actual technology,” said Tuan.