# What I Learned So Far by Lauta > Lessons from running companies in Latin America by Lautaro Schiaffino — founder of Rodati, then Sirena (sold to Zenvia, 2020), now building Darwin AI. Plus a blog, notes, and a few side rooms. This site collects the operating lessons that survived three companies and a decade across LatAm — written as standalone, declarative units rather than a chronological feed. Each lesson is one idea; each company is its own folder. ## About Lautaro - Argentine serial founder, currently CEO of [Darwin AI](https://www.getdarwin.ai) ($7M raised, AI sales agents for LatAm SMBs). - Previously co-founder/CEO of Sirena (acquired by Zenvia, 2020). - Before that: operator at Rodati (vehicle marketplace). - Writes in English, Spanish, and Portuguese. - Twitter/X: [@lautaschiaffino](https://x.com/lautaschiaffino). - Newsletter: [letters.lauta.blog](https://letters.lauta.blog). ## Lessons from Darwin AI AI sales agents for SMBs across LatAm. Currently building (2023–present). - [Build AI from LatAm. Don't apologize for the region — use it.](https://www.lauta.blog/darwin/build-ai-from-latam): The region's constraints aren't a disadvantage. They're an edge — if you build with them in mind. - [Be the Dumb CEO](https://www.lauta.blog/darwin/dumb-ceo): If AI keeps getting better, the dumbest founder in the room — the one who delegates everything — might end up the most effective. - [Most founders think they're tier 1. They're tier 2 at best.](https://www.lauta.blog/darwin/ai-tiers): A three-tier self-assessment for how a person uses AI. Most overestimate themselves. Aim higher. - [If you're over 30, manage your prefrontal cortex](https://www.lauta.blog/darwin/prefrontal-cortex): AI rewards founders who can delegate to non-deterministic systems. The brain region that holds you back finishes developing around 25–30. - [Build for a world where every customer gets their own software](https://www.lauta.blog/darwin/dynamic-software): If AI keeps compounding, the marginal cost of building software approaches zero. That changes everything about B2B product. - [The only AI moats in LatAm are local data and regulation](https://www.lauta.blog/darwin/regulation-and-data-moats): Two questions: will the regional market atomize, and will US winners eat the local ones? Both have the same answer. - [You're not competing against AI. You're competing against the person who uses it well.](https://www.lauta.blog/darwin/future-of-work): Two timelines for thinking about jobs. Both end with: humans do the human things. - [Let humans do human things](https://www.lauta.blog/darwin/let-humans-do-human-things): Darwin's why. The question we have to keep asking is: what is human? - [Build like you're sailing. Never bet against AI.](https://www.lauta.blog/darwin/never-bet-against-ai): Your sail has to be as big as possible, pointed in the right direction, before the wind comes. - [Position yourself in the middle of the meta sandwich](https://www.lauta.blog/darwin/meta-sandwich): Big tech goes after the two slices of bread. The ham, cheese, and tomato in the middle is yours. - [Veto anyone on your team who doesn't believe AI keeps getting better](https://www.lauta.blog/darwin/veto-ai-detractors): A small internal rule that protects belief in the trajectory. - [AI models are gravity fields. Your prompts are masses.](https://www.lauta.blog/darwin/gravity-field): Why the same prompt behaves differently across models, and what to do about it. - [Software is now non-deterministic. Be the human who's responsible.](https://www.lauta.blog/darwin/non-deterministic-software): First time in history a technology takes decisions for us. The human's new job: skin in the game. ## Lessons from Sirena B2B sales-engagement SaaS for Latin America. Sold to Zenvia in 2020. - [Start with the deer. It's the only animal that gives you optionality.](https://www.lauta.blog/sirena/start-with-the-deer): Pick your customer-size animal carefully. You'll spend a decade with it before you can move. - [Run your company on an operating system, not on your mood.](https://www.lauta.blog/sirena/run-your-company-on-an-os): Pick a management methodology — EOS or Scaling Up — and run the company on it. Two reasons most founders skip this; both are wrong. - [Drop CAC and LTV. Look at payback period.](https://www.lauta.blog/sirena/cac-ltv-myth): Why CAC/LTV is noise in early stage, and why payback period is the metric that actually matters. - [Write the angry email. Don't send it.](https://www.lauta.blog/sirena/emails-you-dont-send): Drafting an angry email and not sending it is one of the highest-leverage communication exercises I know. ## Lessons from Rodati A vehicle marketplace in Argentina (~2010s). My first operating role. - [A team that never fights is a broken team.](https://www.lauta.blog/rodati/team-without-conflict-is-broken): Lencioni's Five Dysfunctions of a Team rewired how I build every team. The counterintuitive part: the dysfunction isn't the conflict — it's the absence of it. ## Blog Short essays, things I'm noticing, half-formed thoughts that aren't yet curated lessons. - [Book club](https://www.lauta.blog/blog/book-club): Los pibes lectores: the reading club I started with four friends on Goodreads. - [Am I having enough meetings?](https://www.lauta.blog/blog/enough-meetings): The arrows metaphor and why some bad meetings are a sign you're working right. - [How to hack an AI](https://www.lauta.blog/blog/how-to-hack-an-ai): Prompt injection: how chatbots get broken, and how people try to defend them. - [The networking FOMO curse](https://www.lauta.blog/blog/networking-fomo): How to stop suffering for missing every event — and build something that brings the right people to you instead. - [When is the perfect time to start your company?](https://www.lauta.blog/blog/no-perfect-time): Spoiler: never. And that's the best news. - [Lessons from the Rubik's cube](https://www.lauta.blog/blog/rubik-cube-lessons): What I learned solving cubes works suspiciously well for solving companies. - [5 days to PMF? Spin to win](https://www.lauta.blog/blog/5-days-to-pmf): A five-day sprint for validating product-market fit, adapted from Google's Design Sprint. - [10% of my LinkedIn viewers work in a stealth-mode startup](https://www.lauta.blog/blog/stealth-mode-startups): What stealth mode actually buys you, what it costs, and why the second matters more than the first. - [Investors invest money in people who invest time](https://www.lauta.blog/blog/investors-invest-time): What actually happens when capital meets a founder, and what doesn't. - [The philosophy of this blog](https://www.lauta.blog/blog/blog-philosophy): Why I write here, what the rules are, and what to expect. ## Other pages - [/whoami](https://www.lauta.blog/whoami) — bio, what I'm working on, how to reach me. - [/portfolio](https://www.lauta.blog/portfolio) — startups I've invested in. - [/press](https://www.lauta.blog/press) — interviews, talks, mentions. - [/colophon](https://www.lauta.blog/colophon) — design notes about this site. - [/newsletter](https://www.lauta.blog/newsletter) — subscribe to the Substack. - [/rss.xml](https://www.lauta.blog/rss.xml) — RSS feed of new lessons.