Turning a training methodology into a digital product

A coach who owned a gym in the United States had developed a comprehensive — and quite complex — method for creating personalized workout plans. The method combines health information, a score resulting from movement analysis, and several other individual parameters for each client.

However, all this logic was managed manually across multiple spreadsheets, from data entry to calculation results. Besides making the process difficult to use, this structure hindered standardization and knowledge transfer, preventing other coaches from applying the same method consistently.

The challenge, therefore, was to transform this logic into a structured web system capable of simplifying the creation of workout plans for both the owner and his team.

In parallel, gym members had no digital solution to follow their workouts and exercises individually, creating the need to develop a mobile app for that experience.

The project took place in a context of negotiations with potential investors, which required fast cycles of planning, prototyping, validation, and adaptation. To keep up with this pace, I incorporated AI tools into the Product Design process, using Lovable for the web system and Bolt for the mobile app.

I started by structuring user flows and journeys in medium-fidelity prototypes, using them as a base to define the architecture of the main features and validate the product logic before implementation.

From these prototypes, I used structured prompts in Lovable to progressively generate each part of the web system. This process allowed product and interface decisions to be turned into functional prototypes much faster, keeping control over flows, business rules, and requirements defined during the design process.

For the mobile app, I adopted a similar flow: the first versions of the screens were generated in Bolt, refined in Figma when needed, and later updated again in Bolt. This created a continuous cycle between design → AI generation → refinement → validation → implementation.

Using AI also made the prototyping and iteration process much faster. With functional prototypes available from the early versions, I was able to conduct usability tests with other gym coaches, observe how they interacted with the system, and quickly identify friction points or issues in applying business rules.

The learnings from the tests were then incorporated directly into the product, allowing fast iterations between one validation round and the next, without needing to manually rebuild each change.

This experience was also important for rethinking my own design process: AI began to act not only as an execution tool, but as part of the workflow to accelerate exploration, prototyping, and validation, keeping product decision-making, usability, and solution consistency under design responsibility.