How we take an idea from a first conversation to a working system, and how we use AI in our own engineering.
From idea to working system
01 Discovery
What happens
We map your users, content, systems and constraints with the people who will use it.
What you get
A written specification, architecture and phased estimate. It is yours to keep.
02 Phase plan
What happens
We split the build into phases you can approve one at a time.
What you get
A scope and acceptance criteria for each phase, agreed before work starts.
03 Build
What happens
Short build cycles with regular demos on a test environment you can use.
What you get
Working software early, with release notes for every delivery.
04 Acceptance
What happens
You test against the agreed criteria, and we fix anything that does not meet them.
What you get
Sign-off on clear terms, with a report of every fix.
05 Handover or hosting
What happens
We hand over the code, documentation and access, or keep running and improving the system for you.
What you get
Documentation, an API contract and automated tests, so any team can maintain it.
How we define done
Acceptance criteria in writing
Every feature has written criteria agreed with you before we build it.
Only you mark work as approved
Every item is tracked in a view you share with us. It moves to approved only when you sign it off.
Changes go through a change request
New ideas are welcome. Each one is estimated and agreed before it joins the plan, so scope and budget stay clear.
AI-driven engineering
Our engineers work with AI coding agents every day. It lets a small, senior team take on large systems, with every change documented, tested and reviewed.
AI does more of the typing. People stay accountable for every line that ships.
How every change is made
- A fresh agent session for each task, so the work stays focused
- Many small changes, each easy to review
- An automated security and robustness review of every change
- Tests, documentation and a changelog written with each change
- Every change records what it does, why, how it was checked and its risk
- A person reviews every change before it is merged, and nobody can push straight to the main codebase
Working together
A weekly meeting
A regular call to review progress, see demos and take decisions.
One shared view of the work
Every item, its status and its acceptance criteria in one place.
Direct access to the team
You talk to the people building your system, on the channel you prefer.
Ownership and security. When the system runs on your cloud, you own the code, the data and the accounts. Lumination AI Ltd is Cyber Essentials certified and registered with the UK Information Commissioner’s Office (ZC110820).
Training works the same way. Workshops and AI literacy programmes start with a conversation about your team’s needs and are built around your own tools and examples. See our services.