AI
AI, but with a human at the helm
Work that used to take months can now take days. That part is real, and it changes what is worth attempting. It also means it has never been easier to build the wrong thing very quickly.
Scratch That Itch helps organisations work out where AI genuinely helps, get the thing built, and check it still holds up once the novelty has worn off.
What AI has actually changed
Quite a lot, and not always the things people expect.
- 01
The floor has dropped
Tools, calculators, microsites and internal systems that were priced well beyond small and mid-sized organisations are now within reach. Not cheaper versions. The actual thing.
- 02
The first version arrives immediately
What used to take weeks of discovery and design now appears in front of you in an afternoon. That is genuinely useful, and it is also where some of the trouble starts.
- 03
You can keep going yourself
The most useful outcome is usually not the thing that gets built. It is that your team can see how it was built and carry on independently.
The seven day test
AI output is convincing straight away. That is the whole problem. Something appears in front of you in seconds, fully formed and faster than you expected, and it is very hard not to be impressed by it.
Come back a week later and you often see it differently. The thing that looked finished turns out to be an average first draft. The calculation that looked right was never checked against real data. The copy read well but said the wrong thing.
That is the habit worth getting into. Not whether it looks good today, but whether it will still look good in seven days, once the novelty has gone and real people are relying on it.
Build quickly. Then check it properly, before anyone depends on it.
How we work on an AI project
Four stages, though plenty of projects only need the first one or two.
- 01
Aim
Work out where AI will make the biggest difference, and what is worth building first. The technology can do a great deal, so the useful question is not whether to use it but where to point it.
- 02
Research
Find out whether the idea holds up. What already exists, what it would take, where it is likely to break and what would need to be true for it to be worth doing.
- 03
Build
Where it makes sense, get something real built and in front of people quickly, using whichever tools genuinely fit the job.
- 04
Check
Put it through proper human review before it goes anywhere near your customers, your staff or your data. This is the stage people skip.
Being honest about AI
Nobody has twenty years of experience in this. The tools are months old and there are new ones every week. Some of them are excellent. Plenty of them are not, and the only way to find out is to try them properly on real work.
What does carry over is knowing how to test whether something is any good before committing to it. That has not changed at all, and after more than 25 years of digital projects, it is the part we are most confident about.
If the AI route turns out to be a dog, we will tell you it is a dog. If it can be turned into a greyhound, we will do that, and show you how to train it.
Already built something with AI?
Plenty of organisations now have something that worked beautifully in a demo and is proving harder to live with. We can look at what is there, work out what is salvageable and tell you honestly whether to fix it, finish it or start again.
Discuss a project in difficultySigns worth taking seriously
- It works in a demo and struggles with real data
- Nobody is quite sure how it actually works
- The person who built it has moved on
- It has quietly stopped being used
- You are not confident enough to let customers near it
Ways to start
Most people start with a conversation and go from there.
01· Start here
A Thinking Day
A focused day to work through the idea, look at what already exists and agree whether it is worth taking further.
Typical engagementOne day
02
AI Opportunity Review
A structured, independent look at where AI or automation could create genuine practical value, and what it would take to get there.
Typical engagement2 to 4 weeks
Let’s talk
Think AI could help, but don’t know where to start?
That is the most common place to begin, and it is a perfectly good one. You do not need a defined project, a budget or a particular tool in mind.
Tell us what you have been wondering about, what keeps getting parked, or what you have seen someone else do and thought might work for you. We will help you work out whether it is worth doing.
No obligation and no sales pitch. Just a useful first chat.