04 · AI & automation
AI with a purpose
Practical AI and automation that reduce repetitive work, improve processes and create real value.
Typical timeline: 4 weeks for a first, well-scoped automation
Typical for a well-scoped first release, agreed in week one. Larger scopes are planned in phases.
What it is
AI is useful when it removes work people shouldn’t have to do by hand, and when you can trust what it produces. We build practical automations and assistants around your existing tools, with clear limits, human review where it matters, and costs you can see.
Who it’s for
- Teams spending hours on repetitive requests, documents or data entry.
- Businesses with knowledge spread across documents nobody can search.
- Companies curious about AI who want an honest assessment, not a demo.
Typical scope
We start with one task, measured before and after. A first automation covers the workflow end to end, the integrations it needs, review steps for the people involved, and monitoring. If it doesn’t earn its place, we say so.
How we work
How the project runs
01Week 01
Discover
Understand the business, clarify objectives, identify users and agree on scope.
You receiveA written scope and plan, agreed together.
02Week 02
Design
Shape the experience and settle what will be built before development begins.
You receiveDesigns you can click through, and the technical plan.
03Weeks 03–04
Build
Develop, test, refine and prepare the solution for launch.
You receiveTested software, live on accounts in your name.
04Ongoing
Run
Agreed maintenance, support and improvements after launch.
You receiveUpdates and improvements, or a documented handover.
Four weeks is our standard for a well-scoped project, agreed together in week one — not a promise for every project. Larger work is planned in phases, and ongoing support follows the engagement we agree.
Quality, shown
The technical standard
What every project includes, whichever service you choose.
See how this site measures up| Commitment | Standard |
|---|---|
| Code review | Code review on every change before it’s merged |
| Automated tests | Unit and end-to-end tests, with a DevSecOps pipeline, scope agreed per project |
| Accessibility | Designed and tested against WCAG 2.2 AA |
| Performance | Performance budgets targeting “good” Core Web Vitals, tested on mid-range Android |
| Security | Security review based on the OWASP Top 10 before launch |
| Updates | A written update every week |
| Ownership | Repository and accounts set up in your name; you own the code as set out in your contract |
| Handover | Documentation, architecture notes and a handover walkthrough (recorded on request) |
AI & automation
Frequently asked questions
What kind of work can AI realistically take on?
Repetitive work with clear inputs: sorting and answering routine requests, extracting information from documents, drafting replies for a person to approve, searching your own documentation, moving data between tools. We start where the gain is easy to check.
What happens to our data?
Before go-live, we tell you which AI provider will process your data, in which region, and under which provider terms. Where the provider offers it, we choose settings that keep your data out of model training. Sensitive data can be kept out of the AI system entirely if you prefer. How the provider itself handles data is governed by its own terms, which we share with you.
What if the AI gets something wrong?
It will, sometimes. That’s why a person reviews the output wherever errors would matter, we log what the system does, and we test its accuracy on a sample of your real cases before you rely on it.
Do we need AI at all?
Not always. Some of the best automations are plain rules and integrations. We recommend AI only where it does something simpler tools can’t.