Beyond the hype, AI now offers small organisations a real toolkit: assistants for the writing work, automation for the repetitive work, and meeting tools that take their own notes. A practical starting map.
Somewhere between "AI will change everything" and "AI is just hype" sits the boring, useful truth: there's now a small toolkit of AI capabilities that genuinely help a small organisation, at low or no cost, without a specialist to run them. We've written about AI for one-person businesses elsewhere in the network; this guide is for the next size up — the NGO with six staff, the twenty-person company, the programme office — where the questions become organisational: which tools, for whom, with what rules? What this guide covers: - The four capability areas worth adopting first - What it costs (less than you fear, more than zero) - The rules that keep adoption safe - A 30-day starting plan ## The Four Capabilities Worth Adopting First **Writing and thinking assistants.** The general-purpose assistants handle the organisation's daily word-work: drafting letters, reports, proposals, and policies; summarising long documents; tidying rough notes into usable text. For organisations that produce funder reports or tender responses, this alone repays the effort — not by writing them for you, but by killing the blank page and the formatting hours. **Meeting capture.** Tools that transcribe meetings and draft minutes are among the most reliably valuable AI adoptions for small teams. The minutes nobody had time to write become a by-product of the meeting itself. One rule: everyone in the meeting must know it's being captured — courtesy and, with outsiders, consent. **Document processing.** Turning photographed forms into typed data, extracting details from invoices, summarising stacks of applications — anywhere your team retypes information that already exists on paper, there's now a tool that does the retyping. For programme offices drowning in forms, this is the sleeper hit of the toolkit. **Light automation.** Connecting the systems you already use so information moves without a human courier: form submissions that file themselves, receipts that reach the bookkeeper, reminders that send on schedule. This isn't strictly AI, but it's the same project — removing the repetitive load — and modern automation tools increasingly have AI built in. > The pattern across all four: AI is at its best in a small organisation when it removes work nobody was proud of anyway. Start there — not with the strategy deck. ## What It Costs The honest budget conversation: free tiers of the major assistants are genuinely enough to prove value and build habits. The step up — paid assistant plans for the two or three heaviest users, a meeting-capture tool, maybe an automation subscription — typically lands in the range of a modest monthly software line, not a project budget. What deserves real budget is time: a few hours to set rules, and a champion who helps colleagues past the first awkward week. Tools without adoption are just subscriptions. ## The Rules That Keep You Safe Three rules, set before rollout, save every later headache. **Data:** no personal information about beneficiaries, staff, or clients into consumer AI tools — POPIA applies, and free chatbots aren't contracted data processors. Anonymise, or use tools under proper agreements. **Verification:** AI output is a draft; a named human owns every fact, figure, and commitment before it leaves the building. **Transparency:** the team knows what tools are in use and for what — quiet AI use breeds both risk and resentment. Write the three rules on one page, and you have an AI policy better than most large organisations currently have. ## The Champion Question One predictor separates organisations where AI adoption sticks from those where it fizzles: whether someone owns it. Not a committee, not "everyone" — one named person who's genuinely curious, has permission to spend a few hours a week on it, and likes helping colleagues. The champion's job is small and specific: keep the one-page rules current, collect what's working into shared prompts and shortcuts, sit with the sceptics for twenty minutes when they're stuck, and honestly report what's saving time and what isn't. In a six-person NGO this is a hat someone wears; in a twenty-person company it's a real slice of a role. And measure like you mean it, in units everyone respects: hours. "The monthly funder report used to take two days; it now takes half a day" is adoption evidence a board understands. If after two months nobody can name the hours, the tools aren't failing — the adoption is, and that's fixable. ## A 30-Day Starting Plan Week one: pick one workflow that hurts (funder reports, minutes, form capture) and one tool for it. Weeks two and three: two or three people use it on real work, with the rules above. Week four: review honestly — time saved, quality, problems — and decide: adopt, adjust, or drop. Then pick the next workflow. Adoption by appetite beats adoption by announcement, every time. ## Where Bighearts Online Fits Bighearts Online helps small organisations put technology to work — including choosing AI tools, setti...