From Prompt to MVP at Claude Colima

This Thursday, I’m presenting From Prompt to MVP with Claude at Claude Colima, and I want to take a moment to say how grateful I am for the opportunity.

Claude Colima is part of something bigger. Anthropic and the Claude Communities team are running eight Claude Code meetups across Mexico and the US over five weeks — Monterrey, Ciudad de México, Mérida, Cancún, Colima, Villahermosa, Hermosillo, and a grand finale in San Francisco that will be the first-ever Claude Code meetup 100% en español. What started as a single meetup in CDMX turned into a tour, because developers across Mexico kept asking the same question: ¿cuándo vienen a mi ciudad?

The fact that Colima is on that list matters. This is a small state, but it punches well above its weight: the Universidad de Colima graduates strong software and IT engineers, software companies are setting up shop here, and 90 minutes down the road sits the Port of Manzanillo — Mexico’s largest container port, in the middle of a $3 billion expansion, with AI and digitalization written into its modernization plan. AI adoption in Mexico isn’t just happening in CDMX and Monterrey. It’s happening in ports, in small states, in the places most people aren’t watching. That’s the whole point.

What I’m sharing

My talk is about the workflow I use every day to support clients with brownfield Rails codebases and to start new projects on the right foot. It’s a three-phase loop built around Claude and Rails:

1. MVP Creator — before writing a line of code, Claude helps me produce a research report, a business plan, a brand guide, a technical guide, and a Claude.md. Each one is reviewed and approved before moving on. Scope gets decided here, not in the middle of a sprint.
2. Design & Brand — voice and microcopy on one track, six prioritized UI mocks on the other. Vanilla HTML/CSS first, so the design decisions are deliberate, not whatever the model defaults to.
3. Spec-Driven Development — the approved documents become specs, specs become tasks, and Claude Code executes them with full project context loaded from Recuerd0, the persistent memory system I built, so context survives across sessions and across teammates.

The stack underneath stays boring on purpose: Rails 8, Hotwire, Tailwind 4, Solid Queue, Solid Cache, Minitest. Plus maquina_components for UI and the Maquina generators for the parts every app needs — passwordless auth, multi-tenancy, the things you shouldn’t be rebuilding by hand. I’ve written before about why I bet on Rails and about the path from context engineering to Recuerd0 — the talk pulls those threads together into a working loop.

Brownfield work gets its own tools. The Rails Upgrade Assistant generates step-by-step migration guides for Rails 6.0 through 8.1 — the kind of thing that usually costs weeks of careful reading and trial runs. And the Rails Simplifier reviews existing code through the lens of 37signals patterns and the One Person Framework, pulling overgrown abstractions back toward something a single person can maintain. Both are Claude Code plugins, part of the Maquina AI tools, and both are what I reach for when a client hands me a codebase that’s been carrying technical debt for years.

The thesis is simple: Rails vanilla is enough. Product complexity should come from the user experience, not from the architecture. Claude is the lever that lets one person hold the whole thing in their head and still ship.

The lineup

I’m not the only one presenting in Colima. The full lineup:

• 🎤 Edson Morfin (Sr. ML Engineer) — RAG, agents & guardrails: the pillars of a robust AI app
• 🎤 David Padilla Chávez (Ruby on Rails Contractor) — A brief guide to navigating the AI era
• 🎤 Mario Alberto Chávez — From Prompt to MVP with Claude
• 🎤 Carlos Andrés Robles Hernández (Data Architect @ Globant) — Building an AI Copilot for Ports: Using Claude to Power PortTech in Manzanillo

If you’re in Guadalajara, Colima is right next door. Come through.

Thanks to the Claude Communities team, to Anthropic, and to every person who applied. The demand proved what a lot of us already knew — the next wave of AI builders is coming from Latin America. See you Thursday.

You can find this and my other talks on my speaking page.