The manual work nobody should still be doing

Sound familiar?

• The same information gets typed into a spreadsheet, then a system, then a chat message.
• Someone spends a morning every week assembling a report from four places.
• Work stalls because a person has to remember to move it to the next step.

What you get

• A written map of the workflow as it actually runs — usually the most useful part
• The mechanical steps automated end to end
• Data flowing between the tools you already pay for, instead of through a person
• Alerts when something needs a human, rather than a person checking

When not to bother

Not worth doing for a task that happens twice a year, or one where the rules change every time. Automation pays off on volume and repetition.

How it works

1. 01

   Find where the money leaks

   A conversation, not a questionnaire. I want to know which task eats the most hours and what happens when it is done badly. Often the thing worth building is not the thing you first asked for.

2. 02

   Scope it and price it before I build

   You get the number and what is included in writing, up front. No hourly surprises, and no starting work before we both agree what done looks like.

3. 03

   Ship something small that works

   The first version solves one workflow end to end. You use it with real data before we widen it — that is what stops a project turning into six months of guessing.

4. 04

   Hand it over properly

   Documentation, access, and a walkthrough. If you want me on retainer afterwards that is a choice, not a dependency I engineered in.

Things I've built

• Zykroarts

  A one-page interactive gallery for a digital artist — 54 works, a full-screen viewer, and a commission enquiry funnel.

• Royale Fitness

  Membership and training management for a fitness centre — bookings, members and schedules in one place.

• Bonkend

  A backend builder that generates database schema and API endpoints from a visual editor.