AI Run Coach

The Problem (Why I Built This)

I built this because, as a runner, I was completely underwhelmed by the existing tools. Apps like Strava were just "data buckets." They'd show me my run, but not what to do about it.

When I first started talking to AI, it was even worse. The advice was dangerously generic.

I knew from my own research that running is a deep, science-based field. There are very clear distinctions between a beginner runner, an intermediate runner, and an advanced one.

This was the real problem: "one-size-fits-all" advice is useless. The AI kept giving me beginner advice, and I had to constantly feed it more and more context just to get it to a basic level of usefulness. It was frustrating.

The Strategic Solution

I realized the AI didn't need "better prompts." It needed a "prompt engine."

This was the hard technical problem I wanted to solve. To be useful, the AI had to first understand my level, and then intelligently assemble a specific, complex prompt that pulled from the right data set (e.g., the advanced scientific research) for that exact level.

So, I architected a full-stack AI application that does just that. It's a true expert system that learns, adapts, and—most importantly—builds its logic from a foundation of real, evidence-based rules. I built it all myself, from the database to the front-end.

Capability Showcase

This isn't just a chat window. It's a robust system I built to solve complex, real-world data and context problems.

• A "Prompt Engine" (Adaptive Rules Engine): This is the core of the solution. It's a system I built that thinks. It dynamically assembles its own logic, pulling from science-based rules to ensure its advice is always safe, relevant, and effective for the specific level of the runner.

• Advanced Data Analysis & AI Tooling: I'm a data geek, and I wanted to see the charts. Years ago, well before it was a common feature, I gave the AI its own set of "tools." It can proactively access the database, pull specific sets of performance data, and generate charts for me so we can analyze them together.

• A Self-Aware Confidence System: The AI "knows what it doesn't know." It intelligently adapts its coaching based on the amount and quality of data it has, preventing the dangerous "AI hallucinations" that plague most generic tools.
• True Long-Term Memory & Context: The system analyzes and remembers every single conversation and workout, feeding those insights back into its "brain." It gets smarter and more personalized with every single interaction, just like a real coach.

What This Means For Your Business

I love this project. It was provided to be exceptionally useful to something I love and a very challenging project. The prompt engine also is a solution that many businesses need.

Your company has different levels of employees, different customer segments, different product tiers. Your AI is much less useful if it can't find the right data for the right context.

The same system I built can solve your biggest bottlenecks:

• Smarter Employee Onboarding: An "AI Mentor" that gives your new junior hire the junior playbook, and your new senior manager the senior manager strategy, all from the same system.
• A Sales Team That Actually Scales: An "AI Sales Coach" that doesn't just give generic advice. It accesses your specific sales framework for your specific industry and your specific customer type.
• Scaling Your In-House Experts: I can build a system that captures your best manager's "brain" and unique processes, turning their individual expertise into a scalable, 24/7 resource for the entire team.