Carlo
AI engineer · Co-founder of Ápice HQ · Based in Mexico, working remote (GMT-6)
Systems running for real clients — and the one I use myself.
Simplified flow
Hanami Hair Studio · 2025–2026
One digital system for a hair salon: a site that answers the static questions and sends every call to action into an Instagram DM, where a multi-agent assistant books, reschedules and cancels against the salon's real calendar.
Simplified flow
Internal tool · 2026
The system I use to find work: it pulls remote job listings every 15 minutes, triages them with rules that live in Postgres — not in the workflow tool — and alerts me on Telegram while they're still fresh.
Simplified flow
Internal tool · 2026
Architecture of a multi-tenant platform in development — and how an owner check copied across 42 call sites became named rules that a linter won't let anyone scatter again.
Simplified flow
Ápice HQ · 2026
Two very different businesses under one brand — audiovisual production and digital infrastructure — and a site that sends each visitor down the right funnel before the first call.
Digital fabric catalog with a fabric → variant → photo data model and an image pipeline.
Google Maps scraping with Apify and n8n to build lists for local B2B prospecting.
Three kinds of projects, one standard: they run in production, with real customers.
WhatsApp · Instagram · Web
Assistants that answer questions, qualify leads and book appointments using your real calendar, prices and policies — connected to the systems you already use, not a generic chatbot.
n8n · APIs · Supabase
The tools you already pay for, connected so data moves on its own: lead capture, enrichment, notifications and reporting — with logs and alerts so you know when something breaks.
Next.js · Supabase · Multi-tenant
From a fast site that answers your customers' questions to a multi-tenant product: data model, authentication, permissions and an interface people actually use.
Monitoring isn't a phase at the end. It ships with the first release.
We map the workflow, the systems involved and what "working" means in numbers before anything gets built.
A small working version on your real data, so decisions are made on something you can try — not on a slide.
Error alerts, logs and a periodic QA review are built from day one. Agents fail in quiet ways; you need to see it before your customers do.
See how this played out in the Hanami case study.I keep watching, fixing and improving the system as your business changes.
I'm an Artificial Intelligence Engineering student at UPIIZ-IPN and co-founder of Ápice HQ, where I lead Ápice Infra — the digital infrastructure, automation and AI side of the company.
What I care about is the part of AI that comes after the demo: agents that hold up with real customers, with code where a model shouldn't guess and alerts so failures don't stay silent. I also study the math underneath it — linear algebra, calculus, discrete math — and I like connecting it to what I build. Next, I want to keep building agents and automations for businesses that depend on them every day.
Tell me what's slowing your team down. On a short call we'll work out whether an agent or an automation is the right fix — and if it isn't, I'll say so.
Remote from Mexico · GMT-6 · English & Spanish