CRM

Umana is a multi-tenant SaaS for aestheticians, massage therapists, tattoo artists, dentists, psychologists and any business that works with bookings. Clients book online 24/7, get automatic email reminders, and the business keeps every patient's history, budgets, payments and team tasks in one place.

It runs in Spanish, Portuguese and English, with subscription billing and a 7-day free trial. You can try it at umanacrm.com.

Every screen, database rule and integration in Umana was built with Claude Code. The first working version, with accounts, businesses and a client list, existed on day one. That part is fast now, and it is the least valuable part.

The real work was teaching the product how a clinic actually runs: who books, who confirms, who sees clinical notes, who gets charged (only the business owner, never invited team members), what happens when a patient cancels, and how a budget becomes a treatment and then a payment. AI does not know any of that. I mapped each flow first, then directed the AI to build it, then tested it against real use.

Getting from "it works on my screen" to a product people can pay for took four months of daily iteration: more than 3,800 commits and over 1,200 reviewed changes, each one on its own branch with a preview before it reached production.

Every area of the app went through end-to-end reviews: bookings, budgets, treatments, payments, finances, team and account settings. Because Umana stores health data, security lives in the database itself (row-level rules per business), not in the interface. AI made each change fast. Knowing what to check, and refusing to ship until it was right, is what made it reliable.

• Live product with subscription plans
  Public SaaS with self-serve sign-up, free trial and subscription billing.
• 10+ tools in one place
  Online booking, patient records, budgets, payments, reminders, review requests, team tasks and dashboards.
• Three languages from day one
  Spanish, Portuguese and English across the app and the marketing site.
• Agency output with a fraction of the team
  A product of this scope usually needs designers, frontend and backend developers and QA. Here, one product designer directed the AI through every step.

Read more about why this matters: AI Can Build Your Product. It Can't Understand Your Process.