Starting with Webflow and the OpenAI API
I started Wellness Whizz in September 2024 as a self-funded side project. The original website used Webflow and the OpenAI API. I am a solo founder and I code, but that first version did not need a large application framework to express the idea.
The product is Wellness Whizz, a free AI supplement guide. Its core interaction is a two-minute quiz about goals, diet, sleep and training, followed by a short supplement list with plain-English reasons, study doses and safety notes.
A website was enough to start
Webflow provided the starting point for the site, and the OpenAI API supplied the AI component. The starting point combined a visual website builder with an AI service, keeping the product centered on one useful interaction.
For a small project, the useful question is what a visitor can accomplish. In this case, the interaction has a clear shape: answer a few questions, receive a focused result, and inspect the reasoning behind each suggestion.
Making the result inspectable
Supplement recommendations need more context than a product name. A result should help someone understand why a suggestion appears, what dose a study used, and what cautions deserve attention. A study dose is context, not an instruction that everyone should take that amount.
The design principle behind Wellness Whizz is to keep the explanation close to the recommendation. AI-generated fluency alone does not establish that a claim is correct. Sources and limitations should remain visible parts of the experience.
The product today
The site now includes 150 evidence-based supplement pages, dosage guides, a protein calculator and a photo scanner for a supplement shelf. It is available in English and Spanish. The quiz and resources are accessible without an account or paywall.
The business model is disclosed affiliate links. That disclosure matters alongside the evidence: readers should be able to see both the basis for a recommendation and how the site is funded.
What I would focus on in another small AI product
I would keep the first interaction narrow, make outputs easy to inspect, and decide early how to explain uncertainty. I would also treat the supporting reference pages as part of the product, rather than relying on the quiz to answer everything.
Starting with Webflow and an API is a viable way to put a focused idea on the web. The more interesting work is deciding what the output must explain to be useful.
You can try the quiz at aiww.io. I would welcome feedback on the result explanations and which parts need more clarity.
Written with AI assistance and reviewed for consistency with the project facts.
Top comments (0)