The early findings from this year’s Charity Digital Skills Report provide an interesting read. Based on the first 240 responses, 88% of charities are now using AI in their day-to-day work, up from 76% last year. The gap between small and large organisations has practically closed. The full report isn’t out until July, but even from this early snapshot the picture is clear:The sector has moved fast with AI adoption.
But if you’re thinking “I’m using it, I’m just not sure I’m using it well“… you’re not alone. That’s the conversation we’re hearing everywhere.
The same early data from the Charity Digital Skills Report shows 55% of charities say limited skills are their biggest barrier, up from 43% last year. And concerns about quality, accuracy and ethics are all rising too.
The problem is that most advice out there treats AI as one thing, lumping together everything from learning how to prompt through to transforming your entire organisation. But these are genuinely different challenges needing different skills, and recognising which one you’re actually dealing with is half the battle.
It feels like there are broadly three levels of AI use, and understanding where you sit can help make the whole thing feel less overwhelming.
Everyday AI
This is where most comms teams are, but some may not even realise it.
AI is already baked into many of the tools we use every day, from email software and document editing through to design and social scheduling. Beyond that, people are actively using AI as a thinking partner for drafting, summarising, brainstorming and sense-checking their work.
But even those of us who use large language models (LLMs) like ChatGPT and Claude every day can still end up going round in circles, getting frustrated, only to realise we’d have been quicker doing it ourselves.
The skill needed at this level is developing a feel for what these tools are good at and what they’re not; this helps you know when to step away from their involvement in your work. That judgement only comes through doing, which is why giving yourself (and your team) permission to experiment matters, as well as tapering your expectations of what can be achieved as you and these systems are learning.
There’s real range too, a lot of charities are rolling out tools like Microsoft Copilot across their organisations, which makes sense as a starting point when you’re already in the Microsoft ecosystem. But Copilot isn’t the only option.
Rolling out one tool isn’t the same as having an AI strategy. There are other tools and approaches worth exploring depending on what you’re actually trying to achieve…
Strategic AI
Strategic AI is less about technical AI assistance, like chatbots, and more about asking harder questions. What are the actual problems we’re trying to solve, and could AI, data or digital help us get there faster?
A brilliant AI tool is useless if your data is a mess or your systems don’t talk to each other. (And it’s worth understanding that not all AI works the same way.)
Some systems follow fixed rules and always give you the same answer. Others, like the large language models most of us are familiar with, are more flexible and creative but less predictable. The real power often comes from combining both, alongside your existing technology, to build solutions that genuinely fit the problem.
At Torchbox, we built an AI-powered repository of our previous work so that when we’re preparing a new proposal or marketing piece we can have a conversation with our own data. We can ask something like “how do our design systems support large-scale digital transformations?” and get back an answer drawn from years of project evidence, case studies and thinking, with the right tone of voice and context.
Our delivery team has gone through a similar journey, from struggling to see how AI fitted into day-to-day project work like beyond drafting emails or meeting notes, to more dynamic and strategic solutions, such as building automated sprint reports that pull live data from project tools, querying a shared library of past project learnings to spot risks before they happen, and using AI to turn sales handover documents into structured briefings for incoming teams.
We’re seeing the same approach with charities we work with. Breast Cancer Now, which handles thousands of supporter contacts a year, is exploring AI that drafts responses for specialist staff to review, freeing up time for the work that really needs human judgment.
Art Fund is rolling out an AI-powered tool that helps visitors discover relevant cultural venues based on what they’re actually interested in, a genuinely personalised recommendation service that simply wasn’t possible at scale with a small team. And the DEC, which coordinates emergency fundraising appeals across 15 leading UK aid charities, built an AI proofreader to check content against brand guidelines and appeal-specific wording rules under intense time pressure. What had previously sat entirely in one person’s head could now be codified, shared and scaled. All three started with a specific problem in mind, rather than focusing on the technology they’d use.
This type of strategic consideration needs people who can think about workflows, systems and data, not just prompts. Which raises a question the sector is grappling with: Do you need a dedicated AI team or is it about building these capabilities into existing roles?
Some of the biggest charities are hiring AI specialists and creating new teams. That’s great if you have the scale and the budget. But for most organisations the smarter move might be giving your existing people the time, the tools and the permission to develop these skills.
Transformational AI
This is where AI stops being a faster version of what you already do and enables something genuinely new and innovative.
We’re exploring a project with a large charity that wants to fundamentally rethink how it matches the people it serves with the right support. Instead of relying on manual processes and gut feeling, they’re building an AI-powered matching system that considers dozens of factors at once, ranks the best options and explains why each one is a good fit. That’s not automating an existing process, it’s creating something that simply wasn’t possible to do well at this scale before. And crucially, the intuitive knowledge the team has built about their communities doesn’t disappear in this model. It gets embedded into the system itself, the factors it considers, the way results are weighted, the safeguards built in. That knowledge becomes infrastructure rather than something that lives only in people’s heads.
This is also where new operating models start to emerge. Instead of large teams doing everything manually, a small high-leverage unit (maybe comprising a strategy lead, an engineer and AI working together) can deliver better outcomes with less. This frees people up to focus on the work that matters most.
You don’t need a large budget to think transformationally. For smaller charities, it might look less like a custom-built system and more like fundamentally rethinking a process around what AI now makes possible. The question isn’t “how do we afford this” but “if we were starting from scratch today, how would we do this?”
Start with the problem
The charities who get the most from AI won’t just have people who know how to use the tools, they’ll be asking the best questions.
Is it a content problem, a data problem, a capacity problem or a systems problem? Sometimes the answer is using AI. Sometimes it’s better data, the right digital tools, or simply giving the right people time and space to work differently. Often it’s a combination.
You don’t need to be strong at all three levels equally. A comms officer will lean most heavily on everyday AI. A head of comms needs to think strategically. A CEO needs to understand the transformational picture. But these aren’t hard boundaries — a comms officer who understands the strategic context will always do better work, and a CEO who’s never tried the everyday tools will struggle to lead credibly on this.
Work out where you are, get confident there, and the next step will become more obvious.
The sector is moving fast, and with the right strategy and skillset, you can now do things that weren’t possible even a year ago. The answer isn’t AI for everything. It’s about surfacing the right opportunities and asking better questions.
Useful links
- Read the early findings and take part in this year’s charity digital skill report
- Explore what AI could do for your charity with Torchbox’s AI co-fund
- Use Torchbox’s free AI Policy Builder
AI disclaimer: I used Claude as a thinking partner in the writing process for this article, from shaping the structure to refining the draft. The ideas, perspective and editorial decisions are all mine.
Photo credit: Jackson Sophat on Unsplash
