Claude for Nonprofits.
What you get, what it costs, and how to put it to work

That is the offer. The harder part is what happens after you sign up. We are EdZola, a Coimbatore-based team that builds data systems and AI workflows for nonprofits, and we work as a Claude implementation partner. This guide covers who qualifies, where Claude helps and where it does not, and what India's DPDP Act and UK GDPR mean for you. If you are still working out where AI sits in your wider plans, start with our five-step guide to nonprofit digital transformation.

Anthropic launched Claude for Nonprofits on 2 December 2025 with GivingTuesday, offering up to 75% off Team and Enterprise, three connectors (Blackbaud, Candid, Benevity) and a free course, AI Fluency for Nonprofits. The October 2026 update added:

• nonprofit plans now carry the same models as standard Team and Enterprise plans;
• the Team minimum dropped to two people, and Claude Code is included at no extra cost;
• about 90% off for verified nonprofits in LMICs (announced at the UN General Assembly in September), so Team seats cost US$3;
• a Claude for Nonprofits plugin, installed automatically, with skills for small-team work (marketing, hiring, board), responsible set-up (mapping your organisation, checking your AI policy) and fundraising (finding funders, proposals, reports).

The plugin was built with around 30 design-partner nonprofits; one, Prosper Global, is quoted saying the first version "assumed a small US charity". Expect to adapt templates for FCRA, 80G or Charity Commission language.

Source: claude.com/solutions/nonprofits and Anthropic's getting-started guide. Prices exclude tax and can change; check before you budget.

In money: a 10-person team pays US$80 a month (US$960 a year) at the standard rate, or US$30 a month (US$360 a year) at the LMIC rate. The licence is no longer the expensive part. Staff time to set it up well is.

Who is eligible?

Eligible: registered 501(c)(3) organisations and international equivalents, K-12 schools, and certain US mission-based healthcare providers (Critical Access Hospitals, Rural Emergency Hospitals, FQHCs, Rural Health Clinics). Not eligible: government agencies, political campaigns, universities and large health systems.

Region notes

• India. Registered trusts, societies and Section 8 companies are the likely "international equivalents". India is lower-middle-income per the World Bank, so check for the US$3 rate during Goodstack verification. Keep your registration, 12A/80G documents and PAN ready.
• UK. Registered charities should qualify for the standard rate (US$8) through Goodstack.
• US. 501(c)(3)s verify in minutes. Anthropic's Claude Corps also places 1,000 early-career fellows in at least 400 US nonprofits.

In the 2026 Nonprofit AI Adoption Report by Virtuous and Fundraising.AI (346 nonprofits surveyed in late 2025), 92% said they use AI in some form. Only 7% reported major strategic impact. 81% use it individually and ad hoc; 4% have documented, repeatable workflows. Nearly half have no formal AI policy.

Claude earns its place where work is repetitive, text-heavy and has a clear human checkpoint:

• Grant writing and proposals. Claude turns approved programme material into a structured first draft that fits a funder's template. It is bad at knowing facts you haven't given it, so the quality of your reference pack decides the quality of the draft.
• Reporting to funders and CSR partners. Quarterly donor reports, CSR utilisation narratives, board papers. Claude drafts the narrative from your MIS exports and field notes and flags gaps against the funder's indicators. It should never be the source of the numbers.
• Monitoring and evaluation analysis. Coding open-ended survey responses, summarising themes across transcripts, spotting where quantitative and qualitative findings disagree. IDinsight told Anthropic its teams get surveys field-ready 16 times faster. Keep a researcher who can trace each theme back to the original words; that traceability is the principle behind our work on AI interviews for research.
• Beneficiary and public FAQs. Draft FAQ pages, WhatsApp reply templates and helpline scripts for a programme lead to approve. We do not recommend an unsupervised chatbot in front of beneficiaries on day one.
• Multilingual content. In our experience Claude handles Hindi, Tamil, Spanish, Portuguese and more well enough for first drafts. A fluent staff member reviews anything a beneficiary will read.
• Internal knowledge and onboarding. Policies, SOPs and manuals in a Claude Project become something a new field coordinator can ask questions of. Adoption is quickest here.

If your programme data sits in an MIS, see what AI can realistically do inside an MIS in 2026.

Fix the workflow first. AI makes a broken process faster, not better. Beyond that, these are our hard lines:

• Decisions about people. Selection, eligibility, safeguarding and child-protection decisions stay with trained staff. Claude can summarise a file; it should not decide.
• Raw identifiers. No Aadhaar numbers, health records, caste or religion data, case notes on named children, or donor bank details in a general chat.
• Inventing evidence. No AI-generated quotes, figures or case stories presented as real.
• Final compliance advice. Claude can explain FCRA, 80G or Charity Commission rules; your CA, auditor or lawyer signs off.
• Unreviewed external messages. Anything that leaves the organisation under your name gets a human read.

Anthropic states that by default it does not use Team or Enterprise data to train its models, and says the platform can be configured to support SOC 2 Type II, ISO 27001, ISO 42001 and CSA STAR compliance. That covers the vendor side. It does not cover what your staff choose to upload.

India. The DPDP Rules were notified on 13 November 2025, and most substantive obligations apply from May 2027. You are the data fiduciary, so your consent notice must cover any personal data staff upload; children's data needs extra care; and Claude processing happens outside India, so it belongs on your data map. Our DPDP Act for NGOs plain-English guide explains the law, and the DPDP Rules 2025 compliance checklist turns it into tasks. ZolaShield, our DPDP readiness assessment, ends with a data map where AI tools sit like any other system.

UK. Charities need a lawful basis for personal data processed with Claude, should apply data minimisation, and will often need a Data Protection Impact Assessment where AI use is likely to be high risk. The ICO's guidance on AI and data protection is the reference point.

A one-page AI policy answers six questions: approved tools, what data can go in, who reviews what, when you disclose AI use, who owns it, and what happens when something goes wrong. Then pick two workflows, one low-risk (internal knowledge) and one that matters (grant drafts), build the reference material, connect tools read-only and measure edit effort after four weeks. If nothing worked, the usual reason is reference material, not the model.

Our position is simple: AI sits on top of a connected system of record, with a named person in charge of what it produces. It doesn't replace the system, and it doesn't replace the person.

• Leadership coaching. Our founder works with nonprofit CEOs fortnightly on where AI fits. Most organisations don't need twenty AI tools; they need clarity in two or three places.
• Skills and reference packs for proposals, donor reports and board papers, owned and updated by your team.
• Claude connected to Zoho CRM. Anthropic's ready-made connectors focus on Salesforce, Blackbaud and others. For Indian nonprofits on Zoho CRM or Bigin (often as part of Zoho One for nonprofits), our Amplify cohort is a four-week programme (₹15,000 per organisation, up to two people). Claude reads first, a person approves any record change, and donor emails are saved as drafts.
• Claude over your MIS. On a Zoho Creator MIS, managers ask questions in plain language, with permissions following existing MIS roles.
• Agents with a person in charge. For bounded, high-volume work, agents move through shadow, draft, limited action and managed autonomy. Our AI agents for nonprofits guide explains the method, and DozAI is our product for it.

Our role as a Claude implementation partner. Anthropic's Claude Partner Network (March 2026) has a Partner Hub and public directory for adoption support. Anthropic provides the platform, pricing and training. We help nonprofits, CSR teams and funders, mainly in India and other LMICs, choose workflows, prepare data and policy, connect Claude to Zoho and MIS systems, and train people.

When you probably don't need us. A small team that mainly writes: sign up, take the free course, follow the 30-day plan. A US nonprofit on Salesforce or Blackbaud with in-house IT: Anthropic's connectors and directory may serve you better. Data in a dozen spreadsheets nobody trusts: fix that first. Claude will summarise the mess very fluently, which is worse than not summarising it.