Inspiring Change with AI

Over the past year, Inspiring Scotland and Breeze Digital have been working with four charities through Inspiring Change with AI, a pilot programme exploring how artificial intelligence can help address common operational challenges across the third sector. 

Designed in response to research showing increasing interest in AI alongside widespread uncertainty about how to adopt it responsibly, the programme combines practical support, ethical reflection and shared learning to help organisations explore AI in a way that aligns with their values and priorities.

Having worked through the programme's earlier stages together, each charity has now identified a specific area of their organisation where they believe AI could make a meaningful difference.

In this blog, our programme co-lead, Calum McDonald explores how the charities involved in Inspiring Change with AI are turning ideas into action, examining the practical, ethical and organisational foundations that need to be in place before AI tools can successfully support their work.

From Workshops to Roadmaps

By Calum McDonald, Breeze Digital Associate

Andrew's last two updates set the scene for this pilot well: the original funded charities AI survey told us the third sector was enthusiastic about AI but felt unprepared for it, and the first two workshops turned that finding into a set of shared principles - ask early, start with values, centre people not data - alongside practical tools like customer journey mapping to help each charity see where AI might genuinely help.

My role picks up from there. Since those workshops, I've been sitting down individually with each of the charities, taking the shared thinking from the group sessions and working through what it actually means for their own services, their own teams and their own data. It's a narrower, more practical lens than the workshops: less “what could AI do for the sector” and more “could AI actually be the right tool within our specific context?” and “what does this specific team need to be true before AI helps them?”

I should say upfront that I come at this as an AI skeptic. Wary of the hype, wary of assuming a tool can fix a problem before anyone's properly understood it and concerned with the ethics of the use of AI in the workplace. I think that's turned out to be a useful lens for exactly this stage of the pilot. It gives me a reason to keep coming back to the "ask early" principle I raised back in workshop two: Why this? Who does it serve? Who might it harm?

There is a great deal of excitement around AI, fuelled in part by confident marketing and ambitious claims about what it can achieve. Charities are canny and well-versed in the often anaemic promises of techno-solutionism. The people who work in charities are deeply experienced in thinking about systems and impact already, which really benefitted this pilot project. The main thrust of the conversations moves swiftly away from what AI can do and towards what needs to be true first for any of that to actually help, which is really just the “centre people not data” principle, applied one charity at a time.

A familiar starting point

Talking to all organisations, one thing is common: real enthusiasm to pilot a meaningful approach to AI, paired with opportunities to strengthen data practices to support that enthusiasm.

Charities have built their systems and processes around delivering impactful frontline services with limited time and resource, not around anticipating the specific demands of a new AI tool somewhere down the line. Spreadsheets and CRMs that work well for the team who built them. Data spread across a few different systems reflecting how a service grew over time. Information that's all there, but would benefit from consolidation before it can be confidently used. This isn't a shortfall in good practice, it's simply what “service-focused” looks like for most third sector organisations, AI or no AI. Expecting to find this, and building it into how we approach the pilot, is a large part of why this process exists in the first place.

What it does mean is that “let's use AI for this” is rarely the first sentence in these conversations. The more useful first sentence is usually “let's talk about what we've already got, together.”

This process has allowed us to slow things down and ask questions before we reach for a tool:

●        Who actually owns this data, and would they be comfortable with it going near an AI system?

●        What decision or task are we trying to improve, and is data quality the real blocker, or is it something else entirely?

●        What would it look like if the answer here isn't AI at all?

That last question has been super interesting throughout the process, making it clear that not every problem discussed needs an AI solution. Sometimes the honest answer is a better shared folder structure, or a clearer process, or simply more staff time. Part of what I think this project is proving is that being willing to say "AI isn't the right move here" is what makes the ideas where it could be the right move clearer and more focused.

From conversation to roadmap

This stage of the pilot sits at “Research & Solution Design” moving into “Set Up For Success” - the third and fourth steps of the programme's structure. In practice, that means taking the challenges and ambitions each charity named and going one level deeper with each of them individually. Each of the charities is in a different place, working with different services, different data, and different levels of comfort with technology, but the conversations have followed a similar shape:

• Listen first. Understand the operational pain point in plain language, from the people who live with it daily, not a version filtered through what AI vendors say they can fix.
• Check the foundations. Look honestly at the data and processes underpinning that pain point. Does it support what the organisation is hoping to achieve? Does it need work before progressing?
• Separate the "nice to have" from the "needs fixing." Some ambitions are exciting but premature. Others are modest but ready to go now.
• Build a roadmap, not a purchase order. Rather than landing on a single tool, each charity is coming away with a staged plan. Near-term steps that don't require new technology at all, alongside a longer view of where AI might genuinely help once the groundwork is in place, ready to feed into the programme's next steps.

That roadmap looks different for each organisation. For one, it's mostly about strengthening data foundations before anything else is worth exploring. For another, there's a smaller, well-defined task ready to test now, with the bigger ambitions parked for later. What they share is a sense of sequencing and ongoing discussion, rather than a rush to adopt.

All organisations involved in the programme already bring real care to how they handle the trust placed in them by the people they support. That same care is exactly what's guiding how they're each choosing to approach AI. My role is simply to help make space for that thoughtful approach, and investigate with them how we make progress durable.

In keeping with the programme's commitment to sharing the learning as we go, the next stage of this work is starting to test some of those near-term steps in practice. We will share what we learn, including, I expect, some things that don't go to plan, in a future update.