Gym Plan Drafter: an open-weight AI that drafts safe, budget-aware workout plans for trainers
What I Built Gym Plan Drafter is a small web app for gym trainers. The trainer enters a client's age, job type, goal, experience, days per week, minutes per session, diet type and food budget. In about 10 seconds the app writes a draft weekly workout plan and simple Indian meal ideas. The trainer checks and edits it before the client sees it. Writing a plan by hand takes time, so many clients end…
Gym Plan Drafter is a web application designed to help gym trainers create workout plans for their clients. The trainer inputs information about the client, such as age, job type, goal, experience level, frequency of training sessions, duration per session, diet type, and budget for food expenses. In just a few seconds, the app generates a draft weekly workout plan along with simple Indian meal suggestions.
The trainer then reviews and edits the plan before sharing it with the client. Traditionally, creating a workout plan takes considerable time, and clients often receive identical sheets, which might not cater to specific needs like vegetarian diets or limited time availability. This tool aims to save trainers' time for each new client.
The application is built using GPT-OSS 20B, an open-weight model, accessed through the Groq API. The user interface is created with Python and Streamlit, and it is deployed on Streamlit Community Cloud. The most challenging aspect of developing the app was establishing trust in the model's output, rather than the model itself.
To ensure safety and reliability, the app has two layers of checks: rules set in the prompt and a code checker. The prompt specifies the exact number of days, minutes that add up to the client's training schedule, appropriate food choices based on diet and budget, and the exclusion of calorie numbers. The code checker tests every generated plan against these rules, and if it fails, the app returns the exact issues to the model for another attempt – up to two more iterations. If the plan still doesn't comply, it is displayed with a clear warning.
Privacy concerns were also addressed by not asking clients for body photos, as they pose a significant privacy risk and cannot reliably determine individual needs. The app doesn't include calorie targets, and the minimum age for receiving a plan is 14, while ages 18 and below receive general fitness plans with supervision requirements. The app has undergone 170 automated checks using 10 client profiles to validate the rules and retry logic.
While the app uses an open-weight model, it can easily switch to another open-weight model if needed, without changing the underlying rules and checker. This flexibility ensures that trainers are not locked into a single company's model or pricing. The next step for the developer is to run a small open model directly on the gym's computer, so client health details never leave the premises.
The app has been tested with a real gym trainer who found the generated plan perfect for a fat-loss profile of a 20-year-old MCA student at KNIT Sultanpur on a 4-day schedule. The trainer requested the addition of exercise explanations, which the developer plans to implement in the future. The app's AI usage is disclosed, noting the use of Claude for coding assistance and personal editing of the final post.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
