Arthur Grangeiro de Souza
Seven years building AI inside Latin America's largest bank. I help leaders adopt it with clear eyes: where the value is real, where the risk is, and what to do first.
About Me
For seven years I built machine learning at Itaú Unibanco, the largest bank in Latin America: forecasting, NLP, automation, over £2.6M a year in documented savings. Banks are good teachers on this subject, because nothing ships there without someone asking who is accountable and what happens when the model is wrong. I learned to build AI and to answer for it at the same time. But the project that shaped me most wasn't a model. I founded a data literacy programme that reached 25,000 employees, most of whom would never write a line of code, and it taught me my real job: translation. Most AI problems inside companies are not technical problems. They are understanding problems, and pushback, confusion and stalled strategies are what they look like from the outside. Since then: an MSc in Artificial Intelligence (Distinction), including AI ethics and governance, with hands-on research in generative AI, RAG and multi-agent systems. I now teach AI to professional cohorts every week. Bring me the real situation, not the abstract question. I'll tell you plainly what AI can do for it, what it can't, where the risk lives, and what an honest first step looks like. I have no tool or vendor to sell you, and I never assume knowledge you're supposed to have.
Specializations
Artificial Intelligence, Machine Learning, Analytics
My Strengths
- Built it where it counts. 7+ years of production ML in a regulated bank: £750k/year from forecasting, £1.9M/year from predictive analytics. Responsible deployment learned from the inside, not a checklist. - Translation is my craft. A 25,000-person data literacy programme, a published book, a podcast, weekly teaching. Making AI genuinely understandable is my daily work. - Both sides of the risk conversation. Formal training in AI ethics and governance, plus recent research in generative AI, RAG and multi-agent systems. - I've lived the pushback problem. Adoption inside a 100,000-person organisation taught me that resistance is rarely stubbornness; it's usually a legitimate question nobody answered. - The outsider's eye. First generation into tech from a working-class start. I know how AI looks to the people it's imposed on, which is where adoption quietly fails.
What You'll Gain
- Plain-language clarity on what AI can and can't do for your business, from someone with nothing to sell - One or two concrete next steps you can act on this week - An honest map of your real risks, and which worries you can set down - Framing you can take straight to your board or your team - If your team is resisting: why, and how to turn it into engagement
Topics I Cover
- AI strategy for businesses: where to start, what to pilot, what to ignore - Generative AI and LLMs at work: realistic use cases, limits, and how to evaluate tools - Responsible AI in practice: bias, accountability, transparency, and building trust - AI adoption and change: handling team pushback and building AI literacy across non-technical staff - Explaining AI upward: preparing the board or leadership conversation - Machine learning in practice: forecasting, prediction, NLP, automation, what they cost and what they return
What to Expect
Send me a few lines beforehand, so we spend our minutes thinking rather than briefing. Expect a working conversation, not a lecture: plain language, direct about trade-offs, including the risk side vendors skip. You leave with written-down next steps, even when the honest answer is don't use AI for this. On regulation like the EU AI Act, I'm a practitioner, not a lawyer: I'll show you what it means in practice and when to bring in counsel. 30 minutes for one focused question, 45 to work a situation, 60 to map a strategy. Sessions in English and Portuguese.
Questions & Answers
No, none. Most of my clients are not technical, and making AI understandable to non-experts has been my daily work for years. Bring your situation in your own words; the translation is my job.
Mostly business owners, leaders and professionals who need to make a decision about AI: where to start, what to trust, how to handle risk or team pushback. I also mentor people moving into data and AI careers.
I've spent seven years building AI systems inside the largest bank in Latin America, and just as long explaining them to people meeting AI for the first time. You get the builder and the translator in one conversation, with no tool or vendor to sell you.
Just send me a few lines about your situation and what you'd like to walk away with. That way we spend our time thinking together, not briefing.
Clear, concrete next steps sized to your reality. Sometimes that's a first pilot or a learning plan; sometimes it's the decision not to use AI for something, which can be the most valuable answer of all.
Yes, that's half of what I do. I studied AI ethics and governance formally and built AI in a heavily regulated industry, so we can talk about bias, accountability and trust in plain language. On legal detail like the EU AI Act, I'll help you understand what it means in practice and tell you honestly when to bring in a lawyer.
English and Portuguese. All written follow-ups come in the language of your session.
Schedule your personalized consultation
Pricing
Languages
Book a Consultation
Schedule your personalized consultation