>_ dasalazarr_
Back to Lab
Project 01
Internal Product
Active

AndesRC

AI Coach
WhatsApp
RAG
Training
AndesRC Prototype 1
AndesRC Prototype 2
AndesRC Prototype 3
The Problem

How do you keep a runner from abandoning their training plan? The answer isn't a PDF or another app. It's real accompaniment, in the place where they already are: WhatsApp. The tension is clear: runners want clarity and consistency, but real life breaks plans.

Description

An AI-powered training coach that lives in WhatsApp. The system generates personalized training plans, processes workout feedback in real-time, and adjusts weekly schedules based on actual performance data.

The Solution

A full-stack conversational AI pipeline: TypeScript + Express backend, BuilderBot for WhatsApp orchestration, DeepSeek + GPT-4o mini for intelligence, Postgres (Neon) for structured data, Redis for sessions, and Qdrant for semantic memory. RAG-powered feedback loop that recalculates metrics like ACWR, pace, and rolling mileage after every logged run.

Explore the deeper context

Detailed analysis and implementation results.