AI-assisted building2 min read

How I Use AI to Build and Ship Digital Products

I use AI tools to build and ship digital products, and I direct the work. I define the product, make the decisions, review every change and decide what ships. The tools write code, draft copy and produce variations, and they do it quickly. The accountability for what reaches users stays with me.

It helps to be precise about my background. My core profession is product design, with more than eleven years of work on complex products, including fintech, open banking and B2B platforms. I build my own products with AI-assisted engineering. I do not present that as years of professional software engineering, and I do not expect it to replace an engineering team on a large product. It is a way to test and launch smaller products quickly.

Where AI helps most in my work

StageHow AI helpsWhat I still decide
ExplorationGenerates options for a flow or screenWhich option fits the user and the business
PrototypingTurns a sketch into a working interfaceWhether the interaction feels right
ImplementationWrites first drafts of components and logicArchitecture, naming and review
CopyProduces variationsTone, accuracy and Arabic phrasing
TestingDrafts test casesWhat counts as done

Where I keep control

Authentication, payments, permissions, personal data and anything that affects money stay under close review. I read the generated code in those areas, test the failure cases and ask why each check is there. A demo that works is not enough for those parts.

Product decisions also stay with me. AI tools will happily build a feature that nobody needs. Before I build, I define the user problem, the smallest release that tests it and the measure that will show whether it worked.

How a typical cycle runs

  1. I write a short specification with the user, the core path and the constraints
  2. I ask the tools for one bounded change at a time
  3. I run the tests and check the screen in the browser
  4. I read the diff before accepting it, and reject changes that add complexity without reason
  5. I record what I learned in the product notes so the next session starts from the same facts

What the results look like

The speed gain is real for prototyping and for first drafts. It is smaller for work that requires judgement about users, regulated rules or long-term architecture. Honest tracking matters here, so I measure the time to first working version separately from the time to a version I would stand behind.

What I would tell another designer

Start with a small product you can finish. Write down your decisions before you prompt. Review everything. Keep the human decisions visible in your notes. AI makes building faster, but it does not decide what is worth building.

Next step

Get help improving conversion

Send me the page or flow that loses people. I will tell you where the friction is and what to change first.

Open the conversion page