From "here's what works" to "here's it working"
What build work actually does
Most processes are a chain of tasks. Before AI, every task in that chain was done by a person. After AI, each task can be handled differently depending on what it needs.
Delegate
Some tasks can be handed off entirely. AI completes them, you verify the result. Think: transcribing survey responses, or categorising incoming enquiries against your existing taxonomy. The human role shifts from doing to checking.
Frame
Some tasks work best as a collaboration. You write the brief, AI drafts, you review and refine - iterating until it's right. Think: drafting supporter communications, or exploring a dataset for patterns you hadn't considered. You stay in control of the direction; AI handles the volume.
Keep human
Some tasks should stay entirely with people. Relationship-building, sensitive judgement calls, the work where context and empathy are everything. AI doesn't touch these. But because the tasks around them are now faster, your team has more time and headspace for the work that actually needs them.
Build work is about looking at your processes, figuring out which mode each task belongs in, and then making that real in software. The result is the same work, done faster and at greater scale, with human attention focused where it matters most.
What we build
We don't have fixed product tiers for build work because every organisation's processes are different. But the work tends to fall into a few areas.
Validation and prototyping
Before committing budget, test whether an idea actually works. We build functional prototypes quickly and cheaply, so you can make investment decisions based on evidence rather than speculation. Sometimes the most valuable outcome is learning that an idea isn't worth pursuing before you've spent six figures finding out.
Hearing Dogs for Deaf People
Three revenue concepts prototyped and tested in three weeks. None proved viable for full implementation. The charity avoided an expensive mistake and gained strategic clarity about where to focus instead.
Custom AI tools
Purpose-built tools that solve a specific problem in your organisation. These are designed around how your team actually works, handling the messy reality of your data, your forms, your processes.
Breast Cancer Now
A dual-schema AI system that transcribes complex patient surveys across 20+ different NHS form structures. The four-person team can now work with considerably more hospitals without additional headcount. They named it AIDA.
AI products
Full product development for tools that serve your beneficiaries or your sector. This is the deepest work we do: research, exploration, multiple prototypes, user testing, and iteration until we find something that properly serves the need.
Goose (Arts Marketing Association / National Lottery Heritage Fund)
An AI assistant for heritage marketing professionals. We explored four fundamentally different technical approaches before landing on "thinking partners", which are AI personas that help isolated professionals think strategically through diverse perspectives. Now in use by 40 heritage organisations across the UK.
Why now
The conversation about building custom AI tools is different from even eighteen months ago.
Code is dramatically cheaper
The cost of building custom software has dropped by an order of magnitude. Processes that weren't worth automating at £150k are now viable at £15k. The threshold for what's worth building has shifted, and many charities haven't caught up with where that threshold now sits.
AI tools are capable enough to be useful
Early AI was impressive in demos and disappointing in practice. The current generation of models can handle the messy, unstructured, context-dependent work that charities actually deal with. They're not perfect (and we'll be honest about where they're not) but they're good enough for real work.
Your team already knows where to start
If you've been through training, whether with us or elsewhere, you've already identified the bottlenecks and the places where capacity limits your ambition. The hard part of knowing what to build is already done. The question is execution.
Can't we just do this ourselves?
The tools are impressive. Copilot writes code. Lovable builds interfaces. Claude generates working prototypes from a paragraph of text. The promise from Microsoft, Anthropic, and others is that anyone can build software now. It's natural to wonder whether you need anyone else at all.
You can get surprisingly far on your own. But getting far and getting somewhere useful aren't the same thing.
When the web first appeared, every organisation had a nephew who could make a website. Many of those websites got made. Most of them didn't serve the organisation well, because knowing how to use the tools isn't the same as knowing what to build, how to structure it, or how to make it work for the people who actually need it. We're at that same moment with AI.
The hard part was never writing the code. It's knowing which problems are worth solving in software. It's designing something that fits how your team actually works. It's handling messy real-world data, testing with real users, and iterating based on what you learn. Those are design and engineering judgements, and they matter more now because the cost of building the wrong thing has dropped but the number of things you could build has exploded.
We don't think you need an agency for everything. Some things your team can and should do themselves, and we'll help you see which is which. But for the work that touches your mission, your beneficiaries, or your data at scale, the question isn't "can we build it?" but "will what we build actually work?"
What drives cost
Build projects range widely because the work varies widely. What drives cost is straightforward:
Scope
How many processes or tasks are involved? A single workflow is simpler than reimagining a whole service area.
Complexity
Off-the-shelf tools with some configuration sit at one end. Custom AI with multiple specialised components sits at the other. Most projects fall somewhere in between.
Duration
A three-week validation sprint is different from a six-month product build. We structure work in phases so you can make decisions at each stage about whether to continue, pivot, or stop.
Our recent build projects have ranged from around £8k for focused prototyping to larger investments for full product development. We're happy to give you a realistic estimate once we understand what you're trying to achieve.
How we work
Problem first, not technology first
The most common mistake in AI adoption is starting with "we should use AI" rather than "we need to solve this." We figure out what's actually worth building before we build anything.
Loops, not waterfalls
Each cycle is a focused period of making followed by deliberate learning: testing what we've built, measuring what works, and using those insights to guide what comes next. Now that AI has made the making dramatically easier, the real value is in the learning: human imagination, judgement, and purpose. You see working outputs early and often, and you can change direction before you've over-invested.
Start small, build big
No project begins with a six-month plan and a large invoice. We break complex challenges into smaller components, prove value at each stage, and scale what works. Every phase ends with a decision point: continue, change direction, or stop.
Honest about trade-offs
AI is good at some things and terrible at others. We'll tell you when a spreadsheet is the right answer, when off-the-shelf tools will do, and when you need something custom. We'd rather you spend less with us and get a better outcome.
Built for your team, not ours
Everything we build is designed to be owned and operated by your people. We're not interested in creating dependency. The goal is tools your team can use confidently, with documentation they can actually follow.
Ready to build?
If you know where AI can help and you're ready to make it real, let's talk. We'll give you an honest assessment of what's worth building, what it would involve, and what it would cost.
Technological change continues to accelerate but only a quarter of charities say they feel prepared to respond to the opportunities and challenges. Let's close the opportunity gap together.
