What everyone needs
Give every staff member a paid plan on one of the three major assistants: Claude from Anthropic, ChatGPT from OpenAI, or Gemini from Google. If you take one thing from this guide, take that.
AI free tier plans often lack powerful features that the paid plans have, and may also limit the amount your staff can use AI (e.g. Claude) or limit their access to the smartest models (ChatGPT, Gemini). They also often have worse privacy safeguards.
If you have no reason to prefer one, we recommend Claude. The best model changes every few months and is almost always one of these three, so making the choice beats agonizing over it. Put everyone in the same place, though. Shared tooling is what lets staff swap prompts, compare notes, and help each other.
Check nonprofit pricing before you budget. Claude for Nonprofits, OpenAI, and Google all run programs well below list price, and Google Workspace for Nonprofits includes Gemini at no cost. Our FAQ on what AI actually costs for a small organization has the current picture.
Your organization also needs an AI meeting recorder. The benefits are bigger than they first look. Nobody has to split their attention between listening and taking notes. Decisions and action items stop depending on whoever remembered to write them down. Anyone who missed the call can catch up without asking a colleague to reconstruct it for them, and months later you can search what was actually agreed instead of arguing about it.
Our how-to on turning meeting notes into action items covers what to do with the output, and a resource on choosing a meeting recorder is coming to the library.
Specialists get specialist tools
General assistants serve everyone. Some roles also need something built for their specific work:
- Researchers often want a tool that searches and summarizes academic literature, such as Elicit.
- Designers and creatives may need image generation and editing, such as Freepik.
- Programmers will almost certainly need an AI coding tool. Claude Code, OpenAI Codex, Cursor, and GitHub Copilot are the common choices, and our FAQ on Claude, Claude Code, and Cowork explains how the coding tools differ from the chat app.
Work this out with the person doing the job. They know which part of their week is slow; you do not. Ask what they spend the most time on and what they would hand off first, then look for a tool that fits that answer.
Meet staff where they already work
Asana has AI features. HubSpot has AI features. Gmail has AI features. Whatever your team already opens every morning probably has something built in, and those features cost nothing extra and require no new login.
Quality varies. Teams we work with have found Asana’s sidebar assistant useful; several designers have told us Canva’s AI features are hit or miss. Try them before you assume either way.
The bigger reason to start here is attention. Many staff are already stretched across five or six apps and have spent years building the way they work in each one. Anyone who has watched a spreadsheet expert move through a workbook knows how much of that habit is invisible. Learning a new tool means stepping away from real work to feel incompetent for a while, and plenty of people would rather not. An AI feature inside a familiar tool skips that whole problem.
Power users need power tools
A few people on your team will take to this quickly. They will read about AI in the evenings, arrive with something surprising, and show teammates what they built. Support them.
Being that person is not cheap. Getting to the impressive results usually means hitting the ceiling of a standard plan. The heavier individual tiers from the major providers cost several times a standard seat, coding tools can run to hundreds of dollars a month in usage for someone working at that pace, and a presentation tool like Gamma stops being a novelty for someone building tailored decks for ten meetings a week.
Spending that much on one person’s tooling feels wrong at first. Consider it anyway, because the ceiling is high:
- At an organization that runs institutional plant-based campaigns, a part-time staff member using AI coding tools delivered what the team estimates would otherwise have needed three full-time data scientists.
- An animal advocacy research team moved everyone onto a heavier Claude tier and estimates projects now finish 20–30% faster, which more than covers the subscriptions.
Both figures are the organizations’ own estimates rather than measured studies, and both depend on people who had already put in the hours to get good. Set against the cost of hiring, though, even a fraction of that return pays for the seat. The pillar on tools in our six pillars of AI transformation covers how this fits the rest of your adoption work.
Every tool has to earn its place
The failure mode at the other end is buying eight AI tools and announcing them all in one all-staff email.
A typical worker already moves between a calendar, an inbox, a project tracker, a chat app, and a document editor before lunch. Each new tool adds a login, a set of conventions, a place where information can hide, and one more thing to check. Pile AI subscriptions on top of that and use goes down while resentment goes up, in exactly the people you most need on side.
Before you add a tool, ask who will own it six months from now, which existing tool it replaces, and what happens to the data you put into it. Our guide to data security and privacy basics covers the last one. If a tool has no clear owner and no work waiting for it, leave it.
A rough starting allocation
| Who | What they need | Roughly |
|---|---|---|
| Everyone | A paid seat on one major assistant | Nonprofit rates start around $8 per person per month |
| The organization | One AI meeting recorder | Usually per-seat or per-hour; free tiers exist |
| Specialist roles | One tool for their specific work | Varies by role; researchers, designers, and programmers first |
| Everyone, again | The AI features in software you already pay for | No extra cost |
| Two or three power users | A heavier tier and a real usage budget | Several times a standard seat |
