What we found after running AI readiness assessments with nonprofits across BC, and a practical path to adopt AI safely.
AI has moved from a novelty to a normal part of the workday for a lot of nonprofit staff, often without anyone deciding it should. Someone drafts a grant paragraph with ChatGPT before a deadline. Someone else summarizes a lengthy report before a board meeting. It happens quietly, one task at a time, and it rarely goes through a formal decision at the leadership level.
That quiet, ad hoc pattern is exactly why nonprofit leaders keep asking the same two questions: what’s actually useful here, and what’s safe. Those are good questions, and they deserve a straightforward answer grounded in how nonprofits actually operate.
Watch the full video here
The problem: most nonprofits are already using AI, just without a plan
Across the sector, demand for services keeps climbing while staffing stays flat. That gap in capacity is what pushes staff toward tools like AI in the first place: a way to keep pace without adding headcount.
Tecnet has run AI discovery sessions with nonprofit organizations across British Columbia, and a pattern shows up almost every time we survey staff. About 30 percent say they are aware of their organization’s AI policy, even in cases where leadership is confident that policy has been communicated. When we look at actual usage, closer to 80 percent of staff are already using AI tools on a regular basis. Some are using ten or more different tools, with no organizational visibility into any of it.
This pattern shows up almost everywhere we look, and it reflects how quickly generative AI tools became part of everyday work, faster than most policies could keep pace with. AI is already part of how your team operates today. The useful question is how to support that use responsibly, particularly in an environment where trust, privacy, and mission integrity carry real weight.
The insight: understanding what AI can and cannot do is what makes it safe to use
Workplace AI, in practical terms, is generative technology trained to draft text, summarize content, and organize information based on patterns it has learned. Used well, it handles routine, administrative work: writing an email, summarizing a document, helping someone find information faster. The work itself stays familiar. AI adds to the skill set your team already has.
Its limits matter just as much as its usefulness. AI has no understanding of your organization’s mission or the human impact behind your programs. It cannot make judgment calls, weigh ethics, or carry accountability for a decision. Those responsibilities stay with your staff, every time, whether the task is a two-line email or a public campaign message.
For a nonprofit, that boundary carries extra weight, because the work itself runs on real human connection. The value AI adds here is time. When a staff member spends less time on a recurring administrative task, that time returns to program delivery, donor relationships, and the community the organization serves. As AI use grows across the sector, protecting genuine human interaction becomes even more important.
The framework: match the tool to the task, and the risk to your data
Not every AI tool is built for the same job. A tool that is strong at writing and summarizing can be a poor fit for image editing, and a tool built to automate a workflow may be the wrong choice for open research. Three things determine whether an AI tool produces something useful: the right tool for the task, reliable data behind it, and a clear sense of purpose going in. When any of those three is off, the output tends to be off too.
This is also where the choice between a free and a paid tool becomes a safety question, on top of a budget one. A free or public AI tool typically offers little visibility into where your data is processed, how long it is retained, or who can access it. That’s a real risk when you are working with donor records, beneficiary information, or anything covered by a privacy or funding obligation. Paid and enterprise tools, such as Microsoft Copilot, are built with access control and data residency in mind, and they let your organization see where information goes and who can reach it. A free tool is a reasonable choice for open research using only public information. Anything touching your organization’s own data calls for the enterprise version.
Once a tool and a task are matched to the right level of risk, adoption tends to follow a predictable path. Most nonprofits move through three levels.
Most organizations stay at Level 1 or 2 for quite a while, and that’s a reasonable place to build from as staff gain comfort and confidence. Level 3 makes sense once your data is organized and your team has a track record of reviewing AI output carefully. The right level for your organization depends on your own capacity and risk tolerance.
The solution: start small, protect sensitive data, and build from real use cases
A realistic starting point looks smaller than most of the excitement around AI would suggest. Begin with drafting and brainstorming: first drafts of emails, an outline for a document, a structure for a grant proposal. Keep sensitive information out of these early uses, and keep a person reviewing every output before it goes anywhere. Producing mistakes is part of how AI works, and a human review step is what catches them.
From there, a handful of use cases come up again and again with the nonprofits we work with:
- Research. Gathering background on similar programs, sector trends, or a new initiative before it starts.
- Campaign support. Drafting core messaging, building a campaign calendar, and producing landing page and social content for a fundraising push.
- Policy and procedure questions. An internal chatbot that answers staff and volunteer questions from your own HR or program policies, so your team spends less time answering the same question by email.
- Storytelling. Pulling from program notes and reports to draft an impact story for your website or LinkedIn page.
- Grant application drafting. Using past proposals and background information to produce a first-pass draft for internal review.
- Volunteer onboarding. Registering new volunteers, delivering training material, and generating basic progress reports.
That last item on the checklist deserves its own mention. Building a short human review step into every AI-assisted campaign or document is what keeps accuracy, privacy, and tone in your control before anything reaches a donor or the public.
The policy and procedure use case is worth a closer look, because it’s a strong example of AI use that stays contained and safe by design. Tecnet built an internal agent trained only on our own HR policies and published it to a Teams channel. Staff can ask about a benefit, a time-off policy, or a performance process, and the agent answers using only the uploaded policy documents. Everyone on the team can reach it, even without an individual AI license, and the HR team spends less time answering the same questions by email. A nonprofit could build the same kind of agent around board policies, program guidelines, or funder requirements.
For organizations ready to move past first steps, Tecnet’s AI Scorecard offers a structured way to see where you actually stand. It assesses four areas: strategy, data, workforce, and technical readiness, and rates your organization across four maturity tiers, from early gaps to a fully strategic use of AI. The result is a clear picture of current readiness and a prioritized set of next steps, paired with a short discovery session to walk through what it means for your organization.
None of this requires a technical team or a large budget to get moving. It requires a clear starting point, a plan for keeping sensitive data in the right tools, and a person reviewing the output before it goes out the door.
Ready to see where your organization stands?
Tecnet’s AI Scorecard gives nonprofits a clear, practical assessment of AI and data readiness across strategy, data, workforce, and technical maturity, paired with a roadmap built for your size and capacity.
For a closer look at how task and autonomous AI agents work in practice, including a real example from Tecnet’s own operations, see our companion piece on AI agents for business teams.