AI Does Not Remove Accountability From Humans | Sariol + Caicedo

This is the first time I've had two guests on at once, and I did it on purpose. Jose Carlos Sariol and Guillermo Caicedo both work at Microsoft, helping companies across Central America and the Caribbean get AI past the great demo and into something people actually use, and they work together pretty much every day. (I can barely keep one guest on topic, and now I've booked two.) They're here as themselves, and the opinions are their own.

They've built a framework they call the four Ps: purpose, people, process and platform. So I asked which one stalls an AI pilot, and they don't agree. Jose Carlos looks at process and platform, and says Guillermo leans toward purpose and people. Then Guillermo corrected my question: a lot of these projects do make it into production, and if nobody adopts them and nobody decided up front how to measure success, two or three months later somebody may just shut them down. (I've spent two years asking rooms full of executives why their AI projects stall, and a guy who studied computer science at El Colegio in Mayagüez fixed my question in one sentence.) Jose Carlos went further, and said if you weren't sure what you were trying to achieve, it never should have become a pilot.

I also asked whether AI finally eats culture, so culture never gets to eat strategy (I don't know, that makes no sense, and I said so out loud). Guillermo brought it back to AI literacy, the conversations every organization needs to be having about what's allowed and what you can and can't share.

I asked what humans should keep doing even after AI gets good at it. Jose Carlos says keep humans in the loop, because the life and death decisions need to be made by a human. Guillermo's answer became the title of this episode: "AI does not remove the accountability from the humans." If you're the lawyer, the doctor, the mechanic or the engineer, you'll be accountable. It won't be the AI.

I also told them about the CEO of an angel investment firm who called me wanting a digital twin of himself, one that could think like him, as the company's very first AI project. (I love you, man. Start with literally 999 other things.) Then I asked Jose Carlos to admit I'd been talking to his avatar the whole time, and he said "No, not yet. I'm working on it, but not yet."

They're speaking together at the Caribbean AI Summit, October 9-10 in San Juan, Puerto Rico, on "Why Your AI Pilot Never Made It to Production." Bring a use case of your own: they plan to pull a few from the audience and walk them through the four Ps. I'll be in the front row, maybe not asking questions for once!

I'm Edward Roske, and this is Asking Good Questions. New conversations with the people figuring out the human side of AI. If one of these questions stuck with you, subscribe and send it to someone who'd take the other side.

LINKS
Caribbean AI Summit (Oct 9-10, San Juan): https://CaribbeanSummit.AI
Jose Carlos on LinkedIn: https://www.linkedin.com/in/jcsariol/
Guillermo on LinkedIn: https://www.linkedin.com/in/gcaicedo/

BOOKS & RESEARCH MENTIONED
Gartner's 2026 survey on how many organizations have scaled AI across multiple business units
McKinsey, The state of AI in 2025 (November 2025), on the gap between adopting AI and getting value from it
"Culture eats strategy for breakfast," pinned on Peter Drucker, though Quote Investigator traces it to a March 2000 headline at Giga Information Group
OKRs (Objectives and Key Results), from Andy Grove at Intel
Upload (Greg Daniels, Prime Video) and Pantheon (AMC), two shows about copying a person into a computer
Also mentioned: GPT-6, Copilot, OpenAI, agentic AI, star schemas, snowflake schemas, El Colegio (UPR Mayagüez)

This episode of Asking Good Questions is brought to you by Caprus.Ai.

Check how we empower the office of the CFO to harness the transformative potential of artificial intelligence (AI) and revolutionize their operations.