From prototype to production: Why building reliable Agentic AI is still so hard | Joana Mesquita

From prototype to production: Why building reliable Agentic AI is still so hard | Joana Mesquita

Send us an email! [https://www.buzzsprout.com/2551837/fan_mail/new] Building an AI agent has never been easier. But getting it to production? That's where most projects quietly die. In this episode of Human X Intelligent, host Madalena Costa sits down with Joana Mesquita, Machine Learning Engineer at Swiss Post and former ML practitioner at Adidas and Feedzai, to explore what fundamentally changes when you move from deterministic software to probabilistic, generative AI systems and why so many teams are unprepared for that shift. Joana has spent years building scalable AI systems, implementing MLOps practices and even developing tools to measure the carbon footprint of machine learning workflows. She's one of the clearest thinkers working at the intersection of AI engineering and responsible development. In this episode, we cover: * Why building a working prototype is easy, but building a reliable agentic system is a completely different challenge * What fundamentally breaks when you move from deterministic to probabilistic, generative systems * Why traditional governance models fail for agentic AI and what needs to replace them * How to embed governance into the product itself * Input, output, data and tool guardrails with practical examples * Why evaluation needs to start on day one (and the data behind why it matters) * The risks and trade-offs of using LLMs as judges and how actually to align them * What breaks in the prototype-to-production transition: data quality, cost, latency and governance * How to move from 'trust me, it looks good' to trust backed by evidence and measurement * How organizations can balance innovation speed with responsible AI development * What sustainable AI scaling actually means, including environmental impact One idea that will stay with you: 'Stop thinking about AI products as only the model. Start thinking about them as a system that learns over time.' Whether you're an ML engineer, a product manager or a technical leader navigating the GenAI transition, this conversation will change how you think about what it actually takes to build AI that works in the real world. Connect with Joana Mesquita: → LinkedIn: https://www.linkedin.com/in/joanamesquita96/ [https://www.linkedin.com/in/joanamesquita96/] → Medium: https://medium.com/@joana.c.mesquita.f [https://medium.com/@joana.c.mesquita.f] Human × Intelligent is a podcast at the intersection of design, AI and human agency. Hosted by Madalena Costa. → humanxintelligent.com [https://humanxintelligent.com] → https://www.instagram.com/humanxintelligent/ [https://www.instagram.com/humanxintelligent/] → https://www.linkedin.com/company/human-x-intelligent/ [https://www.linkedin.com/company/human-x-intelligent/] → https://www.instagram.com/designwithmaddie/ [https://www.instagram.com/designwithmaddie/] → https://www.linkedin.com/in/madalenafigueirasdacosta/ [https://www.linkedin.com/in/madalenafigueirasdacosta/] 📩 Want to be a guest on Human X Intelligent? Reach out to Madalena at madalena@humanxintelligent.com Support the show [https://www.buzzsprout.com/2551837/support] 🎙️ Human × Intelligent - a podcast about trust, transparency and human agency in AI systems, for product designers, PMs and founders building with AI.  🔔 Subscribe so you don't miss the next episode  🌐 humanxintelligent.com  Hosted by Madalena Costa · Senior product designer and AI systems strategist

How to use NotebookLM to do real product research (with the prompts)

Send us an email! [https://www.buzzsprout.com/2551837/fan_mail/new] NotebookLM can do in an afternoon what used to take a research team a week, if you know how to prompt it. In this episode of Human × Intelligent, we walk through a complete 8-step AI-assisted research workflow using a real Spotify UX interview study as our working example. Eight participants, 60-minute sessions and a set of raw transcripts, turned into personas, empathy maps, Jobs to Be Done analysis, How Might We questions, an opportunity matrix and a full synthesis report. Every step includes the exact prompt to paste into NotebookLM. No vague instructions. No, just ask AI to help you. Real prompts, real frameworks, real output. What you'll learn: * How to orient NotebookLM before any analysis begins (and why this matters) * How to build 3 grounded user personas, including a tension map that shows where their needs conflict * How to create empathy maps per persona using actual participant language * How to identify functional, emotional and social Jobs to Be Done and rank those that are most underserved * How to generate and prioritise How Might We questions that open up real solution space * How to build a feature opportunity matrix and effort vs impact quadrant * How to affinity cluster raw insights into a 3-level observation → insight → opportunity hierarchy * How to generate an executive summary, full research report and stakeholder presentation outline All 16 prompts are included in the show notes as a ready-to-use guide. This workflow applies to any qualitative research: user interviews, usability test notes, support tickets, survey responses. If you can put it in a document, NotebookLM can help you make sense of it. Show Notes: * Full prompt guide PDF [https://docs.google.com/document/d/1qC-doq2uulFDuvJouNQeTyhL6y8HpDXpSuxYXDRnfXo/edit?usp=sharing] * NotebookLM [https://notebooklm.google.com/] * Follow Human × Intelligent [https://humanxintelligent.com/] Support the show [https://www.buzzsprout.com/2551837/support] 🎙️ Human × Intelligent - a podcast about trust, transparency and human agency in AI systems, for product designers, PMs and founders building with AI.  🔔 Subscribe so you don't miss the next episode  🌐 humanxintelligent.com  Hosted by Madalena Costa · Senior product designer and AI systems strategist

AI doesn't fix broken products. It amplifies them | Michelle Brito

Send us an email! [https://www.buzzsprout.com/2551837/fan_mail/new] AI won’t fix your product. In many cases, it makes things worse. In this conversation, Michelle Brito explains why most companies are getting AI adoption wrong and what they should be doing instead. From her work at Volkswagen Digital Solutions, Michelle shares practical insights on designing AI-powered products that actually deliver value and not confusion. In this episode, you’ll learn: - When artificial intelligence actually makes sense in a product - Why AI is often used as a shortcut for deeper problems - The difference between AI and simple automation - How to evaluate if your workflows are ready for AI - Why user trust breaks when AI is introduced too early ⚠️ The biggest mistake? Starting with the technology instead of the problem. 💡 Key takeaway: AI doesn’t fix broken systems. It amplifies them. Connect with Michelle Brito: → LinkedIn: https://www.linkedin.com/in/michelle-brito-47342554/ Human × Intelligent is a podcast at the intersection of design, AI and human agency. Hosted by Madalena Costa. →   humanxintelligent.com →   https://www.linkedin.com/company/human-x-intelligent/ →   https://www.instagram.com/humanxintelligent/ →   https://www.instagram.com/designwithmaddie/ →   https://www.linkedin.com/in/madalenafigueirasdacosta/ Guest bio Michelle Brito is a Senior Product Designer at Volkswagen Digital Solutions and a mentor at Ladies that UX Lisbon, with over 15 years of experience spanning journalism, editorial design, and digital product strategy. Based in Lisbon, she currently leads design efforts for B2B search engines and researches the integration of AI-driven solutions within the automotive sector. Her background is uniquely multidisciplinary, combining a Master’s in Communication Sciences with a d.MBA and specialized training in UX/UI design. Throughout her career, which includes work for publishing houses, government agencies and marketing firms, Michelle has focused on bridging the gap between business goals and user needs through benchmarking, usability testing and visual thinking. Chapter timestamps 00:00 – Why companies are asking the wrong AI question 00:42 – Introduction to Michelle Brito 01:15 – Is AI the right starting point? 02:20 – Where companies misuse AI (simple problems, wrong solutions) 03:18 – AI as a shortcut for deeper issues 03:39 – Why organizations rush into AI 04:44 – When AI creates confusion and distrust 05:30 – How to push back on stakeholders 06:14 – How to know when AI actually makes sense 07:27 – Why users don’t adopt AI tools 08:24 – Questions to evaluate AI vs automation 08:40 – AI driven by hype vs real need 09:28 – Real example: AI making simple tasks harder 10:10 – Red flags in AI product decisions 11:03 – Why research still matters (even if it’s “boring”) 12:07 – Responsible AI in regulated environments 13:28 – Who is accountable for AI decisions? 14:48 – What healthy AI adoption looks like inside teams 16:40 – Where to start with AI (the right way) 17:21 – The most overlooked first step 18:16 – Making decisions under pressure 19:21 – AI requires simplification, not complexity 19:53 – Practical advice to avoid AI traps 21:06 – Final thoughts on AI hype vs reality Concepts to explore further: → AI vs Automation → AI as a multiplier (not a fixer) → Problem-first vs technology-first thinking → User trust in AI systems → AI readiness (data, workflows, goals) 👉 Subscribe for more conversations on AI, product design and human-centered technology. Support the show [https://www.buzzsprout.com/2551837/support] 🎙️ Human × Intelligent - a podcast about trust, transparency and human agency in AI systems, for product designers, PMs and founders building with AI.  🔔 Subscribe so you don't miss the next episode  🌐 humanxintelligent.com  Hosted by Madalena Costa · Senior product designer and AI systems strategist

The interface trap: Why your AI adoption is failing (and how to fix it) | Kaisa Martiskainen

Send us an email! [https://www.buzzsprout.com/2551837/fan_mail/new] Is your team actually using AI, or are they just playing with it? In this episode of Human X Intelligent, host Madalena Costa sits down with Kaisa Martiskainen, AI Operations lead, to uncover the hidden gap in corporate AI adoption. While usage metrics might be up, true understanding is often lagging. Kaisa explains why providing access to chatbots isn’t the same as building capability and how the 'Interface Trap' prevents organizations from seeing the real value of AI. In this episode, we explore: * The missing conceptual layer: Why mental models are more important than tool proficiency. * The interface trap: How limiting AI to a chatbot window narrows your strategic vision. * Human learning vs. Machine speed: Why humans need friction and failure to truly 'get' AI. * Predictors vs. knowers: Understanding the three foundational concepts every employee needs before their first prompt. * Beyond surface level: How to transition from "interacting" with AI to 'integrating' it into your organizational DNA. If you’re a leader, manager or individual contributor feeling overwhelmed by the AI hype, this conversation will help you shift from reactive usage to intentional system thinking. Connect with Kaisa Martiskainen: →  LinkedIn: www.linkedin.com/in/kaisamartiskainen [http://www.linkedin.com/in/kaisamartiskainen] →  Substack: https://mamaknowsai.substack.com [https://mamaknowsai.substack.com/] Human × Intelligent is a podcast at the intersection of design, AI and human agency. Hosted by Madalena Costa. →   humanxintelligent.com →   https://www.linkedin.com/company/human-x-intelligent/ →   https://www.instagram.com/humanxintelligent/ →   https://www.instagram.com/designwithmaddie/ →   https://www.linkedin.com/in/madalenafigueirasdacosta/ Guest bio Kaisa works at the intersection of technology and human understanding. She helps organizations and individuals understand how to work with artificial intelligence in practical, thoughtful ways, focusing not just on tools, but on how technology changes the way people think, learn and make decisions Chapter timestamps * 00:00 – Use vs. Understand * 01:09 – Real AI Adoption * 02:49 – AI Mental Models * 04:29 – The Metrics Myth * 05:35 – How Humans Learn AI * 07:21 – 3 Rules of Prompting * 10:06 – The Interface Trap * 12:38 – Access ≠ Capability * 15:47 – AI as a Collaborator * 17:28 – The Teaching Problem * 19:54 – A Learning Challenge * 21:43 – The "Ideal" AI Org * 24:32 – The Best Investment * 25:49 – Where to Learn More * 26:57 – Final Takeaways Concepts to explore further: → AI vs Automation → AI as a multiplier (not a fixer) → Problem-first vs technology-first thinking → User trust in AI systems → AI readiness (data, workflows, goals) 👉 Subscribe for more conversations on AI, product design and human-centered technology. Support the show [https://www.buzzsprout.com/2551837/support] 🎙️ Human × Intelligent - a podcast about trust, transparency and human agency in AI systems, for product designers, PMs and founders building with AI.  🔔 Subscribe so you don't miss the next episode  🌐 humanxintelligent.com  Hosted by Madalena Costa · Senior product designer and AI systems strategist

The future of UX: design that knows you better than you know yourself | Joana Cerejo

Send us an email! [https://www.buzzsprout.com/2551837/fan_mail/new] What does it mean to truly anticipate a user and not just what they'll click next...but what they're trying to become? In Episode 19 of Human × Intelligent, Madalena Costa is joined by Joana Cerejo, design lead, AI product designer and author of the Anticipatory Design Playbook. Together, they explore the real depth of anticipatory design, how behavioral science fits into modern AI product work and why most systems fail not because of bad technology but because of a fundamental misunderstanding of human intent. In this episode: - The three layers of anticipation: needs, behavior and outcomes - Why designing for agency can't be an afterthought - Behavioral science frameworks every AI designer should know - The filter bubble problem and collective manipulation - What the Nest Thermostat gets wrong about resilient design - Why transparency is the foundation of everything Connect with Joana Cerejo: → LinkedIn: https://www.linkedin.com/in/jcerejo/ → Website: https://jcerejo.com/ → The Anticipatory Design Playbook (Amazon): https://www.amazon.es/-/pt/dp/1041079109 → Watch Why Personas Fail AI (And What Works): https://www.youtube.com/watch?v=_7dSuJB6M1o&t=897s Human × Intelligent is a podcast at the intersection of design, AI and human agency. Hosted by Madalena Costa. → humanxintelligent.com → https://www.instagram.com/humanxintelligent/ → https://www.linkedin.com/company/human-x-intelligent/ → https://www.instagram.com/designwithmaddie/ → https://www.linkedin.com/in/madalenafigueirasdacosta/ Guest bio Joana Cerejo is a design lead and AI product designer working at the intersection of user experience, behavioral science, and intelligent systems. With nearly a decade of experience designing AI-powered products across fintech, e-learning, and manufacturing, she specializes in making systems that are human-centered, trustworthy, and ethically grounded. She is the author of the Anticipatory Design Playbook, exploring how AI can move beyond predicting behavior to genuinely supporting people in meaningful, long-term ways. Resources & tools section Frameworks mentioned in this episode: → Prochaska Transtheoretical Model - stages of behavioral change; helps design systems that meet users where they actually are → Fogg Behavior Model - behavior happens when motivation, ability, and prompt align at the same time → Nudge Theory - the right intervention at the right moment can make or break a service Book: The Anticipatory Design Playbook by Joana Cerejo - available on Amazon Concepts to explore further: → Filter bubble effect → Human-in-the-loop design → Foresight/futures thinking methodology → AI literacy and explainability Support the show [https://www.buzzsprout.com/2551837/support] 🎙️ Human × Intelligent - a podcast about trust, transparency and human agency in AI systems, for product designers, PMs and founders building with AI.  🔔 Subscribe so you don't miss the next episode  🌐 humanxintelligent.com  Hosted by Madalena Costa · Senior product designer and AI systems strategist