
Read the interview with co-founder Ai-jen Poo to see how they use Claude for Nonprofits.
Founded 19 years ago, the National Domestic Workers Alliance works for the respect, recognition, and rights of nannies, house cleaners, and home care workers. The organization has built a community of roughly 400,000 workers within a workforce the Economic Policy Institute estimates at 2.2 million people, most of them women and many of them immigrants. Ask Aya, its AI support tool built on Claude, helps to bring the collective wisdom of that movement to a worker facing a hard workplace moment alone.

Read the interview with co-founder Ai-jen Poo to see how they use Claude for Nonprofits.
Read the interview with co-founder Ai-jen Poo to see how they use Claude for Nonprofits.
Read the interview with co-founder Ai-jen Poo to see how they use Claude for Nonprofits.
A domestic worker is usually the only worker in their workplace. There is no HR department, no manager to escalate to, and no coworkers to compare notes with. "You're really on your own to negotiate," said Ai-jen Poo, president and co-founder of the National Domestic Workers Alliance. For decades, domestic work sat outside many of the basic labor protections most workers take for granted, like minimum wage and social security. Some of this has changed, but enforcing those protections, or negotiating for anything beyond the legal minimum, still falls to each individual worker.
The stakes are high and personal. "Every time you're asking for something, you are risking potentially losing your job," Poo said, describing a "wild west" with no clear work agreements and little job security. The cost is paid quietly: workers go in sick rather than ask for paid time off, or miss a child's school event rather than risk the conversation. NDWA had spent decades building a response to this: know-your-rights trainings, sample contracts, one-on-one guidance. But that collective wisdom can’t always reach a worker at the moment they need it most, alone and unsure, often late at night.

Turn limited resources into lasting impact. Generate grant proposals, track program outcomes, and free your team to focus on serving your community.
Turn limited resources into lasting impact. Generate grant proposals, track program outcomes, and free your team to focus on serving your community.
Turn limited resources into lasting impact. Generate grant proposals, track program outcomes, and free your team to focus on serving your community.
Chief Strategy and Impact Officer Alistair Stephenson says that Ask Aya, "turns the collective wisdom of thousands of domestic workers into something a worker can access and hold in her hand." Getting there started not with technology but with the workers. NDWA asked 1,000 of them what problem to solve, and whether AI should be involved at all. "We weren't even sure it was going to be AI in the beginning," Poo said. In 2024, members from the organization's home care council traveled to San Francisco to meet with technology companies, then brought their findings back to a national assembly that ratified a set of core principles. "One principle was simply: AI must enhance and not degrade the quality of work," Stephenson said. “Human-centered care must be at the heart of everything." The principles also required that the tool protect worker privacy, stay opt-in, and remain worker-governed.
Worker governance runs through the whole organization. NDWA runs elected leadership councils for its three main sectors: nannies, cleaners, and home care workers, plus an AI working group that meets with the product team monthly to weigh in on design decisions and content accuracy. The team stresses that the councils are not a feedback afterthought. "My bosses are all domestic workers, and elected from our affiliates," Poo said. "If they say no AI, it's no AI." Workers chose the name and the logo and helped design and vet the product through the same board and council structures NDWA uses for its policy work. Stephenson has a name for the approach: "Dignity-driven AI is the type of AI that serves the people most affected by technology, and where those people aren't treated as an afterthought or a focus group," he said.
Anthropic's public emphasis on safety, Stephenson said, was "our entryway into the conversation." From there, the deciding factor was privacy. For a workforce that includes vulnerable communities, the security of user data was the first thing the team evaluated in choosing a model.
NDWA used Claude to create privacy-conscious features when designing Ask Aya. This included a zero-data retention agreement with Anthropic and NDWA building systems to anonymize and aggregate conversations, allowing NDWA to track themes while respecting its users’ privacy. For Poo, the tradeoff was obvious. Worker organizations are "one of the last institutions that working people actually do trust," she said, "and so it's worth not having that data, because it's worth more than anything."
Beyond privacy, the team evaluated each candidate model provider continuously for accuracy, for a tone that sounds like NDWA's own organizers, and for cultural fluency across English and Spanish. Ask Aya routes queries, evaluations, and monitoring across Claude Opus, Sonnet, and Haiku depending on the complexity of the question, keeping the architecture model-agnostic while managing cost.
The hardest evaluation work is the tool’s behavior in high-stakes moments. The NDWA team built guardrails so Ask Aya does not stray into the unauthorized practice of law or sensitive immigration questions, referring workers instead to attorneys, local affiliates, or NDWA's fully staffed worker experience team. “Ask Aya knows the nuances of these cases,” Liibbe said. “We are legitimately getting in contact with people based on those referrals.” A tool like this can look simple, she said, but "building a durable digital product that's embedded in an organization, connected to the rest of the human-centered ecosystem, and verifiably providing accurate and helpful information is much harder."
In a beta with 35 English and Spanish-speaking domestic workers, 93% applied the advice Ask Aya gave them and 25% negotiated an increase in their pay. Another 61% said they felt more influence over their pay, and 50% felt more prepared to resolve conflicts with an employer. Workers come to Ask Aya with concrete asks: how to request time off for a daughter's birthday, how to draft a raise request, or how to write a message to an employer in English when their first language is Spanish.
Trust shows up in the numbers too. "Almost half of our beta testers came into testing very skeptical of AI," Liibbe said, "and 50% left feeling more comfortable using Ask Aya than any other chatbot. Out of the participants who expressed low trust in AI, 93% took concrete workplace action after using Ask Aya." Ask Aya launched in March 2026 and now has about 2,200 users, with a larger rollout planned. The goal for NDWA is reaching 10% of the 2.2 million domestic workers with Ask Aya.
For NDWA, Ask Aya is a door into something larger than any single negotiation. "There is a level of empowerment that can be achieved individually and supported by technology," Poo said. "But there is a different level of impact that can be achieved in community.” According to Poo, “that is the promise of dignity-driven AI: technology built with workers, governed by workers, and connected to the trusted support systems that help some of the most vulnerable workers in our economy build dignity, security, and voice.”