AI is a tool for scale, not a scaling strategy
AI is changing fast. The fundamentals of scaling aren’t. Many organisations we work with feel pressure to use AI to scale and cut costs, but most already have a model that works without it. Here are three things we want them to remember.
Share Post
1. The fundamentals of scaling haven’t changed
You still need a model that solves a problem, a clear scaling pathway and strategy, a sustainable way to fund it, and the people and operations to deliver it.
AI can help with each of those. It replaces none of them.
People still matter just as much. Scale depends on champions who drive the change, on trusted relationships with the partners and communities you work with, and on good judgement and team buy-in about when and how to scale.
And the context the model operates in still sets the ceiling. Bridgespan’s recent research on African nonprofits shows that how far AI can help for scale depends on connectivity, language coverage, data rules and governments’ readiness to adopt it.
2. Scale is more than reach, and AI touches the 3 dimensions of it unevenly
It’s easy to assume that AI helps you scale because it helps you reach more people. But scale isn’t just about reach. Consider what it means for its other dimensions too:
- Depth: how much difference you make
- Sustainability: whether you or others can keep delivering it
Fast Forward’s 2026 AI for Humanity Report found the clearest gains in depth: 55% of 119 AI-powered nonprofits said AI made personalised services at scale possible.
Reach was less clear-cut. Only 13% of those whose reach grew credited AI alone, and 24% said it played a minimal role or none.
And while AI can certainly bring down costs, it by no means guarantees increased sustainability. Prototypes are inexpensive, but running costs rise with use. Most tools are built on someone else’s models, so pricing and terms can change, which also makes it more difficult to hand over to other doers than a human-delivered intervention would.
To reduce that risk to sustainability, favour open or swappable models, keep ownership of your data, and check exit terms before you sign. Funders can support by paying for running costs, monitoring and testing, not just the pilot.
3. AI allows bolder ambition, but you still need to start small
Boldness is where we start with every organisation we work with. The first step is thinking about the size of the problem and setting an audacious goal for impact.
AI often allows more ambitious goals here. Some things that used to cost too much are becoming affordable, and service delivery can be completely reimagined.
But that doesn’t change the need to start small. Every organisation has to prove that a change works before committing to it. This matters even more with AI, where a tool might look impressive while producing wrong results or costing more to run than it saves.
Starting small also means starting where the risk is lowest. We think of AI adoption as a ladder with three rungs:
- internal operations (such as finance and reporting)
- then tools that support frontline staff (such as help with lesson plans)
- then AI that the people you serve interact with directly (such as a chatbot that answers their questions)
In practice
Start with operations, to allow for relatively safe experimentation without jeopardising the people you serve and to build internal capabilities and buy-in that will allow you to rethink how you deliver.
And involve the people you serve in designing and testing anything that will reach them, including whether it works in their language and context.
Still learning
It’s early days, and while this is the way we’re currently approaching this work with partner organisations, the world is shifting fast.
We’re still thinking about all that AI means, have plenty of questions, and we expect our thinking to change as we learn.
If you see it differently, or you’re finding something that doesn’t fit what we’ve said here, we’d welcome your feedback and challenge.