Nicolas Alonso

DocAgenda is an AI-powered receptionist that answers calls and WhatsApp messages around the clock, checking real-time availability and booking appointments directly into a clinic's calendar.

It automates reminders, confirmations, and rescheduling to cut no-shows and free staff from repetitive phone work. Built for healthcare in Colombia and LATAM, clinics can go live in under 72 hours.

Static Ad Gen is an AI-powered ad creative studio that generates production-ready DTC and SaaS static ads in seconds. Brands input their identity, and the pipeline extracts Brand DNA, then produces on-brand visuals and copy using free AI providers via the OpenRouter stack.

Deployed on GCP Compute and proxied through Cloudflare, it uses Claude for copywriting and Grok Imagine for image generation (8 sec/image). Google OAuth handles auth, with a local demo mode available for testing without credentials.

This tool solves the Capacitated Vehicle Routing Problem (CVRP) — the math behind efficient last-mile delivery and field-service dispatching.

Originally built to route 1,200+ field agents across Bogotá and Medellín under a $110k+ monthly operations budget, replacing a manual spreadsheet process and reducing total daily driving distance by roughly 30%. Upload a CSV of stops and watch optimized routes appear on an interactive map with real road geometry from OSRM, solved in seconds by Google OR-Tools Guided Local Search.

A multi-agent pipeline that generates SEO-optimized web copy using Google ADK. Four specialized AI agents collaborate in sequence — Brand DNA extraction, SEO Research, Copywriting, and Quality Assurance — with an iterative QA loop that rewrites until the content meets a quality threshold.

Supports 12 page types (blog posts, landing pages, service pages, etc.), multilingual output (ES/EN), and real internal-link injection from a brand's sitemap. The QA agent scores content across 8 categories — brand coherence, SEO technical, factual accuracy, humanization — and sends it back for revision if the score falls below 88/110.

An honors-awarded credit risk prediction and loan portfolio management system developed during Correlation One's Data Science for All program.

Analyzed large-scale financial transaction history, built machine learning classifiers (XGBoost, LightGBM) to forecast payment defaults, and designed analytics dashboards.