|

Passive vs active AI, and the cost of institutional drift

AI is moving so fast that what counted as early adoption six months ago already feels like a legacy process. If you’re building an AI plan for the rest of 2026, start by working out where your organization stands today.

I think of it as four stages. Most of the nonprofits I talk to sit in the first two, and that’s fine. You can’t plan a route until you know where you’re starting from.

The four stages of AI adoption

  1. The starting line. You haven’t started yet. You’re following the news, but working out where to begin feels overwhelming.
  2. Copy and paste. You’re using a chatbot. You ask a question, get an answer and paste it into another document. You’re still doing all the manual shovelling between tools.
  3. Integrated. You’re working with something like Copilot Studio or Gemini in Google Workspace. AI sits inside your email and documents and helps with daily tasks.
  4. The frontier. You have dedicated assistants that are always on. They don’t wait for a prompt. They spot gaps and bring ideas to your team as things happen.

In the first three stages the work is linear. A person starts every task, and the AI only moves when someone types. At stage four it works in parallel with your staff.

What active AI looks like

Most AI use today is passive. You ask and it answers. That leaves a person as the bottleneck, because nothing happens until someone thinks to ask.

Active AI watches the information it can see and comes to you with a plan. Two examples:

  • A donor posts something on social media. Instead of waiting for you to ask for an email, the assistant checks the post against your records and suggests a 12-month engagement plan to your director.
  • An assistant reading your meeting notes flags that a decision made in the room contradicts a policy in your staff handbook, while the decision is still fresh.

This is AI that keeps working while you sleep, and it’s fair to ask whether your organization is set up for work moving that fast.

Catching institutional drift

One of the best uses I see for active AI is catching institutional drift. Policies and processes change, and the documentation falls behind. Nobody decides to let it happen. It builds up over months while people are busy.

At the frontier, an assistant can audit your internal documents all the time. It reads your policies, project plans and email to find gaps like these:

  • A new program process that’s missing a safety check.
  • A grant report that doesn’t line up with the original proposal.
  • Different teams applying your data privacy policy in different ways.

It runs in the background, so your leadership team gets the list of gaps without going looking.

Research that runs all the time

For most nonprofits, research is a project that happens once a quarter. Someone looks at trends and social media and writes a report.

An active assistant can read news, social media and legislative updates in close to real time, looking for anything that touches your mission. If a conversation about your cause starts trending, it drafts a rapid-response campaign for your comms team. If the government announces new funding, it checks the announcement against your current program capacity right away.

How to move up a stage

You don’t need to jump to stage four this year. Moving one stage forward is a good goal. Here’s how I’d go about it:

  1. Name your stage honestly. Ask a few staff how they actually use AI day to day. Teams are often spread across two or three stages.
  2. Cut the copy and paste where it hurts most. If you’re at stage two, find the task where people paste AI output into another tool most often. That’s your best candidate for an integrated tool that works inside your email or documents.
  3. Start a drift list. Pick one policy that changed in the last year. In a tool your organization has approved for internal documents, ask AI to compare it against the documents that mention it. Each mismatch is a small fix, and the list shows you how much drift you have.
  4. Set up one watch topic. Choose one area your mission depends on, like a funding program or a piece of legislation. Have AI pull a short weekly summary of what changed.
  5. Decide who acts on what it finds. An assistant that flags gaps is only useful if someone owns the follow-up. Name that person before you turn anything on.

For leaders: get ready to manage AI that acts on its own

The technology for stage four is arriving now. For most nonprofits, the harder part will be managing AI that brings work to you instead of waiting for instructions.

My strategy and training work helps teams move from the linear side, with all the copying and manual tasks, to the parallel side, where AI takes an active part in the work. Before you get there, I’d want answers to these:

  • Where does each of our teams sit on the four stages today?
  • Which of our policies and documents would an AI audit flag first?
  • Who on staff has time to review what an active assistant brings forward?
  • What should an assistant never do without a person signing off?

If you’re not sure where you stand, start with my free Nonprofit AI Readiness Roadmap. Answer nine questions about how you use AI in fundraising, programs and admin, and you’ll get a custom roadmap with quick wins for each.

Similar Posts