How Small Charities Can Use AI With Almost No Budget

Start with the free offers built for charities: Google Workspace for Nonprofits includes the Gemini app and Gemini Notebook at $0, Microsoft gives eligible charities Business Basic with Copilot Chat free, and Canva's nonprofit programme covers up to 50 people. Then pick three time-heavy jobs, keep beneficiary data out of personal accounts, and spend $0 to $20 a month.

The catch is that "free" moves the cost into volunteer time and data risk. The small charity that gets AI wrong rarely overspends. More often, six volunteers each use their own free chatbot account, paste in supporter emails and case notes, and the charity has no idea what is stored where. A no-budget plan is mostly about avoiding that.

Follow me on Instagram@sagnikteaches

Stage 1: Get verified once (one to two weeks)

Most nonprofit offers need proof that you are a registered charity. Technology companies use verification partners, most commonly Goodstack, and some use TechSoup; Google and Microsoft run their own nonprofit programmes. Do this once, with the charity's official email address and registration details to hand, and keep a note of which offers you applied for. Approval can take anything from minutes to a couple of weeks.

Connect on LinkedInSagnik Bhattacharya

Use a charity email domain (yourcharity.org rather than a personal Gmail address) for every account you set up. It is the difference between the charity owning its tools and a volunteer owning them.

Subscribe on YouTube@codingliquids

The full list of discounts, and what each requires, is in AI tool discounts for charities and non-profits. The stages below assume you want to spend nothing to start.

Stage 2: Choose the free workspace you will live in

Pick one of the two big suites, not both. The AI you get at no cost differs:

If your team already works in Word and Excel, stay with Microsoft. If you are starting from nothing, Google's free edition gives you more AI for $0. Gemini Notebook alone is useful for a small charity: upload your policies, past funder reports and trustee minutes, and ask questions that are answered from those documents with references back to them. Microsoft's free Business Premium grant for charities ended on 1 July 2025, so ignore older articles that promise it.

Here is what that looks like in practice (illustrative). The coordinator uploads the volunteer handbook, the lone-working policy and the last two years of trustee minutes to a notebook, then asks: "What does our policy say a volunteer should do if a member doesn't answer the door on a home visit?" The answer quotes the policy's steps and shows which document and section they came from. The first time she tried it, the answer came from a 2023 version of the policy that was still in the folder, which said to phone the coordinator's old mobile number. The fix was housekeeping, not AI: keep only the current version of each policy in the notebook, with the date in the file name. Once that was done, new volunteers could ask the notebook first and phone the coordinator only when the answer wasn't clear.

Stage 3: Pick three jobs that eat the most hours

Take a hypothetical befriending charity for older people: one part-time coordinator on 20 hours a week and about 30 volunteers. A month of the coordinator's time might break down like this (illustrative):

The first three rows add up to 15 hours. If a free tool cuts each in half, that is seven or eight hours a month back, or roughly seven more 15-minute phone calls a week to members. That trade is the whole argument for AI in a small charity: less time writing about the work, more time doing it.

Stage 4: Set one data rule everyone can remember

Long AI policies do not get read by volunteers. One rule, in traffic-light form, does. Put this at the top of your shared prompt document and in the volunteer handbook:

GREEN - fine in the charity's AI tools
Public information, your own writing, event details, anonymised
figures ("42 members visited in March").

AMBER - only in the charity's workspace account, never a personal one
Supporter names and emails for thank-you letters, volunteer rotas,
internal documents, draft funder reports.

RED - never in any AI tool
Members' health, care needs or family situations; safeguarding
concerns; bank details; anything a member told you in confidence.

The line between amber and red is the one that matters. A befriending charity hears a lot about members' health and family; none of that belongs in a chatbot, even a business one, unless your trustees have looked at it properly with a data-protection adviser. The rule also gives volunteers a quick answer when they are unsure, which is when mistakes happen. A fuller version belongs in your small charity AI policy.

Stage 5: Write the prompts once and share them

Instead of everyone improvising, save three prompts, one per job, in a shared document or as Gems in Gemini. Each should include your tone, what to leave out and a sample of your own writing. Here is one for rota messages, with the kind of output it produces (illustrative):

Prompt: You write messages to volunteers at a small befriending charity.
Warm, brief, first names only, no exclamation marks, under 80 words.
Changes this week:
- Tues 2pm visit to member M12 moves to Wed 2pm
- Need cover Thurs 10am, member M07, 45 mins, short walk from bus stop
Write one message to the volunteer group.

Output: Hi all, two changes this week. Tuesday's 2pm visit to M12 has
moved to Wednesday at the same time. Could anyone cover Thursday at
10am for M07? It's about 45 minutes and a short walk from the bus
stop. Reply here and I'll confirm. Thank you, as ever. [Coordinator]

Notice that members are referred to by code, not name, so the prompt stays green. The coordinator adds the member's first name when sending to the one volunteer who takes the visit. One fix was needed on the first try: the model added "You're all amazing!", which the "no exclamation marks" line then removed.

For funder reports, the prompt works best when you paste the funder's own questions and your figures, then ask for a draft in their format, marked with [CHECK] wherever a number appears so you verify each one. An illustrative exchange:

Prompt: Funder question: "What difference has the grant made this
quarter? (max 150 words)". Our figures: 42 members visited weekly,
6 new members, 3 volunteers trained, 1 member moved into care.
Member quote (consent given): "Tuesday is the day I talk to someone."
Draft an answer. Mark every number [CHECK]. Use only these facts.

Output: This quarter our volunteers visited [CHECK] 42 members every
week, including [CHECK] 6 who joined us for the first time. We trained
[CHECK] 3 new volunteers, which let us take on those referrals without
a waiting list. One member told us: "Tuesday is the day I talk to
someone." ...

Two edits before sending: the draft claimed the new volunteers meant "no waiting list", which the coordinator hadn't said and which wasn't quite true, and it left out the member who moved into care, which the funder would want to know about as part of an honest picture. The [CHECK] markers made the numbers quick to confirm.

Social posts follow the same pattern. Before, a volunteer's AI-drafted post read: "We're so incredibly grateful for our amazing volunteers who make a huge difference every single day!" After adding one real detail and asking for plain language: "Last Tuesday, 14 volunteers made 42 visits. For some members, it's the only conversation they'll have all week. If you've got an hour on a weekday, we'd love to hear from you." The second post gives readers a fact, a reason and something to do.

For the donor side, thank-you letters that still feel personal and appeals and social posts that keep your voice go further. Keeping the prompts in one shared document stops each volunteer reinventing them.

Stage 6: Trial it for four weeks with one volunteer champion

Pick one person, ideally a volunteer who is comfortable with technology, to own the trial. Their job: use the three prompts, note the minutes each job took compared with before, and record every correction they had to make. At the end of four weeks, the champion and the coordinator answer three questions:

1. Did each job take noticeably less time, including checking?
2. Were any corrections serious (a wrong fact, a wrong name, something that would have embarrassed the charity)?
3. Did anything amber or red end up where it shouldn't?

If the answers are yes, no and no, extend it to the next job on your list. If a serious error appeared, find out whether the prompt, the data or the checking failed before going further. For a newsletter, a realistic target is a first draft in ten minutes and an edit in twenty, against an evening's work before.

What month one might look like at the befriending charity

Here is a plausible first month for the example charity, to show what the champion's notes should capture (illustrative figures):

• Rota messages: from 6 hours to about 3.5. The saving came from not rewriting the same swap message five times; checking dates still took time.
• Newsletter: from 5 hours to about 2. The draft came in minutes; choosing photos and getting members' permission to feature them did not change.
• Funder report: from 4 hours to about 2. One serious catch: the draft said "over 50 members supported this quarter" when the figure pasted in was 42. The [CHECK] marker on every number is why it was spotted before sending.

That is about seven and a half hours back in the month, at a cost of $0, and one error that proves why the checking step can't be skipped. It is also the kind of summary trustees respond to: hours saved, what they were spent on instead, and what went wrong and how it was caught.

Stage 7: Decide whether anything is worth paying for

After a month, some charities find the free tier is enough. Others hit limits: the free Gemini app won't write inside Gmail, and Copilot Chat doesn't search across your organisation's email and SharePoint files the way the paid Copilot does. The low-cost options for verified charities, as of September 2026:

One detail worth checking: Microsoft's nonprofit price for the Copilot add-on is $25.50, while the commercial Microsoft 365 Copilot Business plan for organisations of up to 300 users lists at $21 a user a month on annual billing, with an $18 promotion to 31 December 2026. A small charity may pay less on the commercial plan. Ask your Microsoft partner or check both in the admin centre before buying.

For most small charities, two seats on a business chat plan for the coordinator and one trusted volunteer ($16 to $20 a month) is the only spend worth considering in the first year.

If you're a community group without charity registration

Many small groups, such as a residents' association, a parent-run club or a mutual aid network, aren't registered charities, so the nonprofit offers above aren't open to them. The plan still works with two changes. Use the free individual tiers of the chat assistants, but set up one shared account under a group email address rather than personal accounts, and switch off use of chats for model training in its privacy settings. Then tighten the data rule: in a free consumer tool, treat amber as red. Names, contact details and anything about members' circumstances stay out entirely; the tool is for public-facing writing, event planning and tidying your own notes. If the group later registers, move to a nonprofit workspace and bring your saved prompts with you. Canva's nonprofit programme also requires registration, so plan around its free tier until then, and keep your templates in the group's own account so they aren't lost if a volunteer moves on.

A one-page AI induction for volunteers

New volunteers need three minutes on AI, not a policy. A filled-in version for the befriending charity:

Using AI at [charity name]
- Use the charity's Google account only. Never your own AI apps
  for charity work.
- Green / amber / red: see the card on the office wall.
  Members' health and family situations are ALWAYS red.
- Saved prompts are in "AI prompts" in the shared drive. Use them
  rather than starting from scratch.
- Every AI draft is checked by a person before it's sent or posted.
  Numbers get checked twice.
- Not sure? Ask the coordinator before pasting anything.

Go through it at induction and ask the volunteer to repeat the red rule back in their own words. It sounds basic; it is also the thing most likely to be forgotten under time pressure.

Where "free" ends up costing a small charity

• Free tiers change without notice. Limits, features and model names in free plans move often. Don't build a process that depends on a specific free allowance.
• Tools disappear. One AI calendar tool, Clockwise, shut in March 2026 and wiped user data rather than passing it on. Keep your prompts, templates and outputs in your own documents, not only inside a tool.
• Volunteers leave. Anything built in a personal account leaves with them. Charity accounts only.
• Checking takes time. If nobody has 20 minutes to read the funder report draft properly, the saving is illusory.
• Trustees find out late. A short note to the board before you start is far easier than an explanation after a problem.

If you would like a second opinion on whether you need outside help at all, do small charities need outside help to adopt AI? gives an honest answer, and turning charity data into impact reports is the natural next job once the first three are running.

Questions small charities ask about free AI

Do our trustees need to approve using AI?

It is wise to get their agreement, even for free tools, because trustees are responsible for how the charity handles data and reputation. A one-page note covering which tools, which jobs, the data rule and who checks the output is usually enough for a board meeting. It also protects the coordinator if something goes wrong later.

Can volunteers use their own ChatGPT or Gemini accounts?

For public tasks such as rewording a social post, the risk is low. For anything involving supporters, beneficiaries or internal documents, no: personal free accounts may keep or train on conversations, and the charity loses the work when a volunteer leaves. Give volunteers access through the charity's own workspace instead, where the data protections apply.

Is AI worth it for a charity with no paid staff?

Often more so, because volunteer time is the scarcest resource. Start with one job that eats a volunteer's evening every month, such as the newsletter or funder report, and see whether a free tool halves it. If nobody has time to check what the AI produces, wait: unchecked AI output is where charities get into trouble.

Further reads

• Best Free AI Tools for Small Business Owners and Their Catches — More free tools, and the catch in each one.
• Shadow AI: How to Stop Staff Pasting Client Data Into Free Tools — Keep volunteers out of personal AI accounts for charity work.
• Charity AI Risks: Data Protection, Deepfakes, and Donor Trust — The data, deepfake and donor-trust risks in more depth.
• Can You Use AI for Grant Writing Without Funders Rejecting It? — How funders view applications drafted with AI help.
• How Charities Use AI Chatbots to Answer Supporter Questions — When a chatbot for supporter questions is worth it.
• Copilot Chat vs Microsoft 365 Copilot: What Do You Get Free? — What the free Copilot Chat does and doesn't include.
• How to Build a Shared Prompt Library for Your Team — How to organise a prompt library a whole team uses.
• How to Use AI to Write Grant Applications, Step by Step — Nine steps from funder guidance to submission: AI extracts criteria, drafts to character limits and plays the assessor. Every figure comes from your evidence.
• AI Grant Writing Tools Compared for Small Non-Profits — Specialist grant tools against general assistants on non-profit pricing: what each costs, what it does well, and which suits 4, 12 or 30 bids a year.
• Using AI to Spot Donors Who Are About to Lapse — A step-by-step method for flagging donors who are drifting, using a pseudonymised export, an overdue-ratio score and a person checking the list.
• Can AI Help a Charity Forecast Demand for Its Services? — Use Excel's Forecast Sheet or a chat assistant to turn two years of service counts into a winter range, then add what the model can't see.
• How to Write a One-Page AI Business Case for a Small Business — Five stages to a one-page AI business case, a finished example from a small charity, and a drafting prompt that stops AI inventing figures.
• What Is the Cheapest Way to Start Using AI in a Small Business? — A $0 first fortnight, the AI features sitting unused in your software, a church office's first month, and when the first $20 is worth paying.
• AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.

Sources: Google for Nonprofits help pages on AI features in Workspace for Nonprofits; Microsoft Nonprofits offers page; OpenAI help pages on OpenAI for Nonprofits; Claude for Nonprofits page; Canva for Nonprofits newsroom and help pages.