Not sure what you need?
Start with a review.
If the requirement is unclear, or comes from a regulation, our advisory team can assess it first and write the specification the software is built to.
We design, build and run custom web applications for organisations of any size and sector, from small firms replacing spreadsheets to large regulated businesses. We add AI where it saves real effort: reading documents, ranking options and guiding assessments.
What we build
Typical projects: an internal tool that replaces a spreadsheet, a customer or supplier portal, document and email processing, reporting, and integrations between systems. We build the whole system: the AI features, the database, user accounts, roles and reporting. People make the final decision, and every AI result shows what it was based on.
Upload a CV, form or contract in any layout, including scanned PDFs and two-column designs. The software returns structured fields (name, skills, dates, certifications) and keeps the original file alongside.
Compare people, suppliers or options against a written specification. Each result shows the matching points, the gaps and the reasons behind its score.
Questionnaires that change with each answer, evidence upload, review and sign-off, user roles, and PDF and Excel reports. The questions live in data files, so a change in the rules is a content update, not a code change.
One application serving many companies, each with a separate database, its own logo and colours, and its own settings. Colours, logo and font can be picked up from the customer's website.
Using AI safely
Every AI feature we build has a person making the decision, a spend limit, bias and hidden-instruction tests, an audit trail and a security scan before release.
How a project runs
We write the requirements as numbered, testable statements, and list the risks of each AI feature.
Each code change is linked to the requirement it implements, and each requirement to the tests that check it.
Before release: security scans, bias and hidden-instruction tests on the AI models, and a scored risk register.
We monitor AI cost per customer, apply database updates to every customer's database, and write a deployment guide.
Built with Python (FastAPI), TypeScript (React and Next.js), PostgreSQL, and Claude from Anthropic for language tasks.
Not sure what you need?
If the requirement is unclear, or comes from a regulation, our advisory team can assess it first and write the specification the software is built to.

Contact
Tell us about the system you want built, or the regulation you need to meet. We will reply with a few questions.