AI-powered bike fitting assistant built on Rails

Enhanced a cycling fit platform with an AI-powered assistant that delivers personalized bike fitting guidance, supports user onboarding, and provides contextual recommendations based on rider data and proprietary fitting expertise.

Smartphone recording a cyclist's riding posture for automated AI-powered analysis in a custom fitness app.

Business context

MyVeloFit is the world’s #1 online bike sizing and fitting platform, helping riders optimize their positioning through AI-powered photo and video analysis. Trusted by individual cyclists and retailers for professional-grade fits, the solution combines mobility checks, riding posture evaluation, and personalized adjustment recommendations through automated, intelligent guidance.

Project in facts

Fitness
Canada
Ruby on Rails developer, PM
April 2024 – May 2024
Ruby on Rails development, AI integration
Ruby on Rails, Hotwire, TimescaleDB, OpenAI

Request project

Challenge

MyVeloFit wanted to enhance its bike fitting platform with conversational AI capabilities that could provide expert-level guidance without requiring direct involvement from human specialists. The goal was to improve user onboarding and reduce support workload.

To achieve this, the client needed an AI-powered chatbot that could analyze rider-submitted data, leverage proprietary bike fitting knowledge, and recommend adjustments based on each cyclist’s body position, mobility, and reported discomfort.

Product goals

Codify proprietary fitting expertise into the product logic to deliver personalized recommendations
Optimize the onboarding flow to increase activation and reduce friction
Reduce support burden through self-service fit validation

Solutions

To ensure domain-specific accuracy, we structured the client’s proprietary bike-fitting documentation, training materials, and business rules into an internal knowledge base to ground the assistant. The core of the solution was an AI-powered assistant that supports cyclists in two key scenarios.

For bike selection, users answer questions about their riding preferences and goals. Based on this information, the assistant recommends a suitable bike.

For bike fitting, the chatbot leverages MyVeloFit’s proprietary knowledge base to provide personalized recommendations for saddle and handlebar adjustments and identify potentially uncomfortable or dangerous riding postures.

We delivered an OpenAI-powered MVP within two weeks. Following user acceptance testing, we scaled the solution by expanding the knowledge base, refining prompt logic, optimizing response speed, and launching a dedicated UI.

Domain-specific knowledge base

Structured proprietary bike-fitting data and business rules for accurate responses.

Selection engine

Combined rider preferences with text-based outputs from the client’s video-analysis workflow to generate personalized bike recommendations.

Posture assessment

Analyzed rider position data to generate personalized saddle and handlebar adjustment recommendations.

Engineering approach

We selected OpenAI for its optimal balance of response quality, latency, API maturity, and operational cost. Our engineers developed the chatbot for RAG‑based responses and integrated it into the existing Ruby on Rails platform. Hotwire enabled interactive chat without the overhead of a separate frontend application.

Conversations were stored within the platform and linked directly to customer accounts, while role-based access controls and secure API communication protected sensitive user data. To efficiently manage growing volumes of conversation history, we used TimescaleDB, whose time-series architecture supports querying, storage, and analysis of chronological interaction data.

AI-powered chatbot guiding users through cycling tips and gear selection

Business impact

AI posture assessment with real-time joint-angle measurements.

Codified domain expertise

Proprietary materials were transformed into a structured knowledge base, allowing our client to deliver consistent recommendations at scale.

Foundation for future AI capabilities

The architecture allows the client to continuously expand the knowledge base and introduce new AI-driven workflows without major changes to the platform.

Result

The expertise of Rubyroid Labs in RoR AI integration services helped MyVeloFit extend its bike fitting platform with AI‑powered guidance, making expert knowledge more accessible and reducing reliance on manual support.

Rapid MVP delivery
An initial version was delivered within two weeks, so the client could validate the concept and iteratively improve the assistant based on user feedback.
Faster user onboarding and self-service support
The chatbot guides users through the platform and helps them find relevant resources, which reduces friction during onboarding and everyday usage.
AI-assisted bike fitting experience
Cyclists receive contextual recommendations without waiting for direct assistance from bike fitting specialists.

Web landing page for a virtual bike fitting service built with Ruby on Rails

Rubyroid Labs was great to work with. They had excellent team dynamics and an extremely quick turnaround.

Jesse Jarjour, CEO, MyVeloFit

Looking to integrate AI into your product?

Please fill out this form, and our manager will contact you within one business hour. If necessary, we can sign an NDA and begin project discussions.

Full name Full name is required Email Email is required Company Company name is required Message Message is required

attach file file size limit - 10 MB

I give my consent for Rubyroid Labs to process my personal data pursuant to Rubyroid Labs Privacy and Cookies Policy in order to handle my request and respond to it. I am aware that I have the right to withdraw my consent at any time.