How we help
From a first idea to something that works
No hype and no hundred-page strategy: a clear view of where AI earns its place in your business, and help making it happen.
Finding where AI fits
A look at how your team works to find the jobs AI is genuinely good at, such as reading and sorting email, summarising documents, drafting replies and pulling details out of messy text, and an honest view of where it isn't worth it.
Prototypes you can try
A small working version, run on your own examples, so you can judge it on real results before committing to anything bigger.
AI built into your software
Features using Anthropic's Claude added to your apps, websites and back-office systems, with structured answers your systems can check and the running costs kept in view.
Automating processes
AI joined up with the services you already use, such as email, Microsoft 365, Google, Shopify or your database, so routine reading, sorting and drafting happens on its own while people approve what matters.
Advice on tools and costs
Help choosing between AI services and plans, what each would cost to run, and the risks to watch for, so decisions are made on the facts rather than the sales pitch.
Data ready for AI
AI is only as good as the data it works from. Two decades of databases and data warehousing, including NHS data, go into getting yours clean, joined up and safely reachable.
How it works
Small steps, each one useful
1. A conversation
Tell us what takes up your team's time and what you'd like to change. A rough idea is enough.
2. A shortlist
We agree the few uses most worth doing, what each would save, and what it would take to build and run.
3. A prototype
The best candidate, working on your own examples, so you can see the results for yourself.
4. Build and support
Built into your systems and looked after for you, or handed over to your own team with notes on how it works.
In practice
AI we've built and use every day
Each one has its own page with screenshots and the full list of features.
JONg Mail
Claude reads only the emails a rule points it at and says whether they need action, alongside rules that file and tidy mail across several mailboxes.
JONg Run
An AI running coach that answers from someone's actual Garmin and Strava training, not generic advice.
Monster Maker
A family project where children build a monster by tapping buttons and AI draws it, with a blocked-words list and a grown-ups page for daily budgets.
How we build
Every app on this site is built with Claude as a development partner: Joe sets the direction, reviews the work and tests it.
Used safely
Careful with your data and your customers
Only the data it needs
AI is given just what a task needs, personal data is left out where it isn't needed, and UK GDPR is part of the design from the start.
Checked, not trusted blindly
Answers come back in a structure the software checks, and people approve the actions that matter before anything is sent or changed.
Your keys, your accounts
Keys are kept as secrets on the server, services run under your own accounts where you want them to, and everything is tested before it goes live.
Tools
What we work with
- Claude and the Claude API
- Claude Code
- Structured outputs and tool use
- TypeScript and Node.js
- SQL Server
- PostgreSQL
- Microsoft 365 and Google APIs
- Shopify
- AWS
- Fly.io
- AI image generation
Wondering where AI could help?
Tell us a little about your business and what takes up your team's time, and we'll reply by email.